Home

Sun Network QDR InfiniBand Gateway Switch Administration Guide

image

Contents

1. I Topology matching results I a E Tt ee ee ee a ee ee ee a E E S Administering the InfiniBand Fabric 99 I The topology defined in ib_topology topo perfectly matches the discovered fabric FabMan gateway_name gt Related Information a Gateway Reference generatetopology command m Gateway Reference matchtopology command m Display the InfiniBand Fabric Topology Simple on page 86 V Find 1x SDR or DDR Links in the Fabric You can use the ibdiagnet command to determine which links are at 1x bandwidth 2 5 Gbps or 5 Gbps data rate On the management controller type FabMan gateway_name gt ibdiagnet lw 4x ls 10 pc pm skip all In this instance of the ibdiagnet command there is a check for all links to be at 4x QDR lw 4x ls 10 and if not to report those links that are not 4x QDR Related Information m Gateway Reference indiagnet command a Display the Link Status of a Node on page 88 m Determine Which Links Are Experiencing Significant Errors on page 100 V Determine Which Links Are Experiencing Significant Errors You can use the ibdiagnet command to determine which links are experiencing symbol errors and recovery errors by injecting packets 1 On the management controller type FabMan gateway_name gt ibdiagnet c 500 P all 1 In this instance of the ibdiagnet command 500 test packets are injecte
2. The script begins Interactive menus direct you through the installation process During the installation two configuration files are created m ofed conf contains the names of the software modules installed and the configuration settings chosen during the installation m ofed_net conf contains the IPoIB configuration settings chosen during the installation The script finishes This information is found in the respective locations a Man pages are installed in usr share man Documentation is installed under the usr share doc directory IPoIB configuration information is installed under the etc sysconfig network directory m The openibd daemon is installed under the etc init d directory m BXOFED commands are located in the usr bin and usr sbin directories a BXOFED software installation information is displayed with the etc infiniband info script 4 Optional If the tgz file was extracted to a NFS shared directory for a cluster then to install the BKOFED software onto any remaining nodes in that cluster a Log in as superuser of a node to receive the BXOFED software b Change to the directory where the tgz file was extracted c Install the BKOFED software automatically BXOFED 1 5 1 1 6 3 install pl c path ofed conf n path ofed_net conf where path is the directory path to the ofed conf and ofed_net conf files d Repeat from Step a for all nodes to receive the BXOFED software
3. Administering the Chassis 21 Measured 1 8V 1 80 V Measured 1 2V Standby 1 20 V All voltages OK FabMan gateway_name gt Related Information m Gateway Reference checkvoltages command a Display Power Supply Status on page 20 m Display Gateway Environmental and Operational Data on page 23 Vv Display Internal Temperatures If you are concerned with the internal temperature of the gateway you can use the showtemps command to display nominal and measured temperatures If there is a great deviation there might not be enough cooling On the management controller type FabMan gateway_name gt showtemps Back temperature 29 Front temperature 29 SP temperature 47 Switch temperature 40 maxtemperature 42 Bridge 0 temperature 40 maxtemperature 41 Bridge 1 temperature 45 maxtemperature 46 All temperatures OK FabMan gateway_name gt Related Information m Gateway Reference showtemps command m Display Fan Status on page 22 a Display Gateway Environmental and Operational Data on page 23 Vv Display Fan Status The output of the getfanspeed command can help you determine fan speed inconsistencies which might indicate a future failure 22 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 On the management controller type FabMan gateway_name gt getfanspeed Fan 0 not present Fan 1 running at rpm 11212
4. FabMan gateway_name gt smsubnetprotection commit FabMan gateway_name gt The configuration is distributed to the trusted standby Subnet Managers and the current secret M_Key to the InfiniBand Fabric Related Information m Gateway Reference smsubnetprotection command m Create the Secret M_Key Configuration on page 67 m Set the Secret M_Key to Use on page 72 m Delete Secret M_Keys From the Configuration on page 73 m Configure a Local Secret M_Key on page 75 Administering the Subnet Manager 71 V Set the Secret M_Key to Use 1 Initiate a secret M_Key configuration session on the management controller that hosts the master Subnet Manager FabMan gateway_name gt smsubnetprotection start FabMan gateway_name gt 2 Display the active secret M_Key configuration See Display the Secret M_Key Configuration on page 52 3 Determine which M_Key you will set as the new current 4 Set the current M_Key FabMan gateway_name gt smsubnetprotection set current m_key where m_key is the management key 16 hexadecimal digits For example FabMan gateway_name gt smsubnetprotection set current 0x00abcdefabcdef02 Mkey 0x00abcdefabcdef02 updated FabMan gateway_name gt Note You can only set an M_Key which is in the configuration 5 Display the pending M_Keys FabMan gafeway_name gt smsubnetprotection list modified File _format_version_number
5. m guid is the global unique identifier of the target port on the host associated with the VNIC This value is column two of the gateway MAC address list m hostname is the host name of the management controller or system name recognized by the BridgeX manager m port is the port of the host m string is the character string in the node description field m mac is the machine address code This value is column three of the gateway MAC address list Note Only even numbers are supported for the most significant byte of the MAC address unicast Note If you do not assign a MAC address the VNIC is not utilized You must later assign a MAC address to the VNIC with the modifyvnic command if you want to use the VNIC a vlan_ID is the VLAN identifier This value is column four of the gateway MAC address list a p_key is the partition identifier default This value is column five of the gateway MAC address list For example to create a persistent VNIC for connector 1A ETH 3 associated with VLAN 3 with partition key default type FabMan gateway_name gt createvnic 1A ETH 3 guid 00 03 BA 00 01 00 E3 71 mac 02 02 02 02 02 01 vlan 3 pkey default FabMan gateway_name gt Note If the VNIC is to be non persistent append the n option to the command line Note For VNICs created by the createvnic command the Network Interface names at the host which is set by the DEVICE parameter in the ifcfg e
6. passwd root Changing password for user root New UNIX password new_password Retype new UNIX password new_password passwd all authentication tokens updated successfully Related Information m Gateway Installation accessing the management controller m Restart the Management Controller on page 29 30 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Administering the 14 Switch Chip These topics describe the administration of the I4 switch chip Monitoring the 14 Switch Chip on page 31 Controlling the 14 Switch Chip on page 40 Related Information Troubleshooting the Gateway on page 1 Understanding Administrative Commands on page 13 Administering the Chassis on page 19 Administering Gateway Resources on page 137 Administering the InfiniBand Fabric on page 83 Administering the Subnet Manager on page 49 Monitoring the 14 Switch Chip These topics enable you to display and check the operation and status of the 14 switch chip Note To use all the commands described in these topics you must access them through the SYS Fabric_Mgmt Linux shell target of the Oracle ILOM CLI Display the Switch Chip Port to QSFP Connector Mapping on page 32 Locate a InfiniBand Device and Connector From the GUID and Port on page 33 Locate a Switch Chip or Connector From the GUID and Port on
7. 1 2 Identify the VNIC you want to modify See Display the VNICs on page 176 On the management controller type modifyvnic connector vnic ID mac mac vlan vlan_ID where Administering Gateway Resources 197 m connector is the name of the connector OA ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH a vnic_ID is the VNIC identifier m mac is the machine address code m vlan_ID is the VLAN identifier NO or 2 to 4094 For example to modify VNIC 15 on connector 1A ETH 1 so that it has a different MAC address type modifyvnic 1A ETH 1 15 mac 02 02 02 02 02 88 VNIC ID 15 modified 3 Repeat from Step 1 for all VNICs to be modified 4 Verify that the VNICs have been modified showvnics ID STATE FLG IOA_GUID NODE IID MAC VLN PKEY GW 16 UP N 00 21 28 00 01 3E CE A0 mnm34 55 0000 02 02 02 02 02 04 3 9234 1A ETH 4 13 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 01 3 9234 1A ETH 3 15 UP N 00 21 28 00 01 3E CE 9F mnm34 55 0000 02 02 02 02 02 88 3 9234 1A ETH 1 14 UP N 00 03 BA 00 01 00 E3 72 mnm34 60 0000 02 02 02 02 02 02 3 9234 1A ETH 2 Related Information m Gateway Reference modifyvnic command m Gateway Reference showvnics command a VNIC Allocation on the Gateway on page 191 m Create VLANs on page 192 a Delete VLANs on page 194 m Create VNICs on page 195 a Enable VNICs on
8. 3 Display the active secret M_Key configuration See Display the Secret M_Key Configuration on page 52 4 Add an M_Key to the configuration FabMan gateway_name gt smsubnetprotection add m_key where m_key is the management key 16 hexadecimal digits Note There is a maximum of 10 secret M_Keys for the configuration For example FabMan gateway_name gt smsubnetprotection add 0x00abcdefabcdef01 Mkey added FabMan gateway_name gt smsubnetprotection add 0x00abcdefabcdef02 Mkey added FabMan gateway_name gt 5 Select the M_Key to use FabMan gateway_name gt smsubnetprotection set current m_key where m_key is the management key 16 hexadecimal digits For example FabMan gateway_name gt smsubnetprotection set current 0x00abcdefabcdef01 Mkey 0x00abcdefabcdef01 updated FabMan gateway_name gt Note You can only set an M_Key that has been added to the configuration 68 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 6 Display the pending M_Keys FabMan gateway_name gt smsubnetprotection list modified File format_version_number 1 Sun DCS IB mkey config file This file is generated do not edit secretmkey enabled nodeid o04nm2 gw 6 time 15 Sep 03 38 33 checksum 0 commit_number 2 Mkey Untrusted Mkey Smkey Attribute 0x00abcdefabcdef01 Oxlaa45124fee612ae Ox15fc26aea300f831 0x00abcdefabcdef02 0x4ccd8230de6c
9. See Enable the Subnet Manager on page 62 If you installed a new switch or gateway attach the signal and InfiniBand cables now Refer to Switch Installation or Gateway Installation Prepare the new Subnet Manager for M_Key functionality See Prepare for Secret M_Key Functionality on page 66 Optional change the priority of the new Subnet Manager See Set the Subnet Manager Priority on page 56 Note If you are setting the new Subnet Manager as the master Subnet Manager you must set the older master Subnet Manager to a lower priority Related Information Merge Subnets Into One Secret M_Key Fabric on page 79 Secret M_Key and Firmware Upgrade and Downgrade Guidelines on page 80 V Merge Subnets Into One Secret M_Key Fabric You might want to combine two or more subnets into a larger fabric with secret M_Key functionality 1 If the firmware on the switches and gateways are less than version 2 1 upgrade them Refer to Switch Remote Management or Gateway Remote Management upgrading the firmware Configure identical secret M_Key lists on each subnet See Prepare for Secret M_Key Functionality on page 66 and Create the Secret M_Key Configuration on page 67 Administering the Subnet Manager 79 Note The smnodes list must have all smnodes of the future fabric and not just those of the individual subnets Additionally the subnets cannot communicate with each other because t
10. e fabricMgmtFabricName Related Information Gateway Reference SUN FABRIC MIB mib Gateway Reference fAconfig command Display the InfiniBand Fabric Configuration on page 94 fdconfig Command Overview on page 126 Create a Fabric Configuration on page 127 Add or Remove an Element From the Fabric Configuration on page 129 Modify an Element of the Fabric Configuration on page 131 dconfig Command Overview The f dconfig command is used to configure a list of fabric elements expected to be in the fabric the fabric node list The list is used by the Fabric Directors and provides these attributes for each fabric element Name The host name of the fabric element as configured by Oracle ILOM IP address The out of band management network IP address of the management controller hosting the fabric element and Fabric Director Fabric element type The name of the type of fabric element sw36 Sun Datacenter InfiniBand Switch 36 sw36gw Sun Network QDR InfiniBand Gateway Switch a unknown The element s type is unknown Role type The purpose of the fabric element gateway The element acts as a gateway 126 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 spineSwitch The element acts as a spine switch a leafSwitch The element acts as a leaf switch unknown The element s role is unknown m Maste
11. on page 192 m Create VNICs on page 195 m Enable VNICs on page 199 m Disable VNICs on page 200 m Delete VNICs on page 201 m Display the VLANs on page 174 Create VNICs You must create VLANs before VNICs All VNICs are assigned to a VLAN even if the VLAN ID is No For consistency you can assign the VLAN ID of No to the P_Key of default 1 If unknown determine the HCA port GUIDs MAC addresses the partition keys and the connectors that will be assigned to the VNICs See Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 2 If a VLAN will be associated with the VNICs create the VLANs before the VNICs See Create VLANs on page 192 3 Optional Use the gateway MAC address list you created earlier See Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 or Determine VLAN Associations for Gateway Manual Mode Linux on page 145 4 Create the VNIC on the management controller FabMan gateway_name gt createvnic connector lagname lt guid guid gt lt host hostname port port gt lt node string port port gt mac mac vlan vlan_ID pkey p_key where connector is the name of the connector OA ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH This value is column one of the gateway MAC address list m Jagname is the identifier of the LAG Administering Gateway Resources 195
12. 2004 LinkUp LinkWidthSupported 1X or 4X LinkWidthEnabled 1X or 4X LinkWidthActive 4X LinkSpee dSupported assen 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedEnabled 0 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedActive 2005 10 0 Gbps Peer PortInfo Port info Lid 15 DR path slid 15 dlid 65535 0 23 FabMan gateway_name gt LINnKkState ceca Rae a ee es Active PhysLinkState 2008 LinkUp LinkWidthSupported 1X or 4X LinkWidthEnabled 1X or 4X LinkWidthActive 0 4X LinkSpee Supported sess sse 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedEnabled 4 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedActive 2 2 eee eee 10 0 Gbps In the output the Active parameters are the current state of the port Related Information m Gateway Reference ipportstate command m Display Switch Chip Port Status on page 37 a Display Link Status on page 35 a Find 1x SDR or DDR Links in the Fabric on page 100 m Set Port Speed on page 103 Display Counters for a Node To help ascertain the health of a node the perfquery command displays the performance error and data counters for that node Administering the InfiniBand Fabric 89 On the management controller type FabMan gateway_name gt perfquery lid port where m lid is the LID o
13. 3 Enable the Subnet Manager See Enable the Subnet Manager on page 62 Related Information m Gateway Reference setsubnetprefix command m Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 m Set the Subnet Manager Priority on page 56 m Enable Subnet Manager Controlled Handover on page 58 a Set the Subnet Manager Readable Management Key on page 59 m Clear the Subnet Manager Readable Management Key on page 60 m Set the Subnet Manager Routing Algorithm on page 61 m Enable the Subnet Manager on page 62 a Disable the Subnet Manager on page 63 V Enable Subnet Manager Controlled Handover If your InfiniBand fabric has two or more Subnet Managers you can force a constrained fallback protocol should the master Subnet Manager fail Refer to Gateway Reference setcontrolledhandover command for more information 1 On the management controller disable the Subnet Manager See Disable the Subnet Manager on page 63 2 Enable controlled handover FabMan gateway_name gt setcontrolledhandover TRUE Current SM settings smpriority 5 controlled_handover TRUE 58 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 subnet_prefix Oxabbababe M_Key None Routing engine FatTree FabMan gateway_name gt 3 Enable the Subnet Manager See Enable the Su
14. Display the VNICs on page 176 a Display IO Adapters on page 177 a Display Gateway Ethernet Port Information on page 178 m Display the Gateway Instance Number on page 181 m Display Information About the Gateway on page 182 Vv Display the Gateway Instance Number Gateways use instance numbers to identify themselves to each other Administering Gateway Resources 181 1 On the management controller type FabMan gateway_name gt setgwinstance list Current GW instance 34 FabMan gateway_name gt 2 If the information displayed is incorrect modify the gateway instance See Set the Gateway Instance Number on page 207 Related Information Gateway Reference setgwinstance command Set the Gateway Instance Number on page 207 Display LAG Information on page 173 Display the VLANs on page 174 Display the VNICs on page 176 Display IO Adapters on page 177 Display Gateway Ethernet Port Information on page 178 Display Gateway Port Information on page 179 Display Information About the Gateway on page 182 Display Information About the Gateway You can display version status and operational information about the gateway with the showgwconfig command 1 On the management controller type FabMan gateway_name gt showgwconfig BXM pid 13724 is running BXM versions bxm_user 2 0 0898 0 BXM API 1 6 0 bxm_libs 2 0 0898 0 b
15. Enable LAG Mode on page 184 Create LAGs Create LAGs on page 185 14 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Step Functionality Description Links 6 Enable host Acquire BXOFED software Acquire the BXOFED Software Linux on support of the page 138 gateway Install BXOFED software Install the BKOFED Software Linux on page 139 7 Determine the Use Gateway Manual Mode Gateway Manual Mode Overview Linux on VNIC mode or Host Manual Mode page 143 Host Manual Mode Overview Linux on page 149 8 Configure VNICs Gateway Manual Create the MAC address file Determine VNIC Configuration Parameters For Mode Gateway Manual Mode Linux on page 144 Append the MAC address Determine VLAN Associations for Gateway file Manual Mode Linux on page 145 Create VLANs Create VLANs on page 192 Create VNICs Create VNICs on page 195 Host Manual Mode Determine the configuration Central Configuration File Linux on page 150 file VNIC Specific Configuration File Linux on page 151 Create the MAC address file Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 Create the configuration file Create the Central Configuration File Linux on page 154 Create the VNIC Specific Configuration Files Linux on page 156 Enable VLANs and VNICs Configure and Create VNIC
16. Si ce logiciel ou la documentation qui l accompagne est conc d sous licence au Gouvernement des Etats Unis ou toute entit qui d livre la licence de ce logiciel ou l utilise pour le compte du Gouvernement des Etats Unis la notice suivante s applique U S GOVERNMENT END USERS Oracle programs including any operating system integrated software any programs installed on the hardware and or documentation delivered to U S Government end users are commercial computer software pursuant to the applicable Federal Acquisition Regulation and agency specific supplemental regulations As such use duplication disclosure modification and adaptation of the programs including any operating system integrated software any programs installed on the hardware and or documentation shall be subject to license terms and license restrictions applicable to the programs No other rights are granted to the U S Government Ce logiciel ou mat riel a t d velopp pour un usage g n ral dans le cadre d applications de gestion des informations Ce logiciel ou mat riel n est pas con u ni n est destin tre utilis dans des applications risque notamment dans des applications pouvant causer des dommages corporels Si vous utilisez ce logiciel ou mat riel dans le cadre d applications dangereuses il est de votre responsabilit de prendre toutes les mesures de secours de sauvegarde de redondance et autres mesures n cessaires a son utilis
17. m Set the Gateway System Name on page 208 a Display Gateway Ethernet Port Information on page 178 a Display Gateway Port Information on page 179 V Set the Gateway Service Level You can set the Ethernet data traffic and control traffic service levels with the setgwsl command Note When enabling EoIB consider that there are conditions from partitioning See smpartition Command Guidelines on page 107 On the management controller type FabMan gateway_name gt setgwsl eoib ctrl leve where level is the service level Note Use service level 1 only for low latency high priority small message low bandwidth traffic Use other service levels for regular high bandwidth traffic For example to set the data traffic service level to 2 type FabMan gateway_name gt setgwsl eoib 2 Stopping Bridge Manager OK Starting Bridge Manager OK FabMan gateway_name gt Related Information m Gateway Reference setgwsl1 command a Enable a Gateway Ethernet Port on page 203 a Disable a Gateway Ethernet Port on page 204 206 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 m Set the Gateway Ethernet Port Parameters on page 205 m Set the Gateway Instance Number on page 207 m Set the Gateway System Name on page 208 a Display Information About the Gateway on page 182 Set the Gateway Instance Number
18. on page 58 Configure the M_Key Set the Subnet Manager Readable Management Key on page 59 Set the routing algorithm if Set the Subnet Manager Routing Algorithm on not Fat Tree page 61 2 Configure the Create a fabric configuration Create a Fabric Configuration on page 127 InfiniBand fabric 3 Secure fabric Create the smnodes list Create the smnodes List on page 133 management optional Configure IPOIB Create the IPoIB Interface on page 135 functionality Create the fabric element Create a Fabric Element Configuration on configuration page 120 Set the replication password Set the Replication Password on page 76 Enable M_Key persistence Enable or Disable Secret M_Key Persistence on page 76 Configure secret M_Keys Configuring Secret M_Keys on page 67 Enable secret M_Keys Enable or Disable Secret M_Key Functionality on page 77 4 Partition the Collect GUIDs Identify All Switches in the Fabric on page 84 InfiniBand fabric Identify All CAs in the Fabric on page 85 Display Gateway Port Information on page 179 Update the smno es file Update the smnodes List on page 134 Create the partition Determine the Partitions and P_Keys on page 109 information file Configure partitions and Create a User Partition on page 110 P_Keys Add GUIDs to the partitions Add or Remove a Port From a Partition on page 112 5 Configure LAGs Enable LAG mode
19. 140 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 5 Reboot the Linux InfiniBand host s Related Information Gateway Remote Management upgrading the gateway firmware Acquire the BKOFED Software Linux on page 138 Acquire the ConnectX 2 Firmware For your host to properly interface with the gateway the firmware of the ConnectX 2 chip in the HCA must be updated to version 2 7 000 or higher 1 2 Open a web browser on the host that will receive the ConnectX 2 firmware Go to this URL http support oracle com Oracle s My Oracle Support page is displayed Sign in if you already have an account The dashboard page is displayed Note If you do not have an account you must register 10 Click the Patches amp Updates tab The Patches amp Updates page is displayed In the Patch Search window click the Product or Family Advanced The Patch Search window updates In the Product Is drop down menu select your HCA For example Sun Dual Port 4x QDR InfiniBand IB HCA PCle ExpressModule In the Release Is drop down menu select the latest firmware version For example FW25408 v2 7 8130 Click outside of the drop down menu In the Platform Is drop down menu select the Oracle Solaris appropriate for your host For example Oracle Solaris on x86 64 64 bit Click outside of the drop down menu Administe
20. 77 list active 97 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 list modified 52 set current 72 setlocalsecretmkey 75 setreplicationpassword 76 version 27 configuration overview 13 configuring Fabric Director node list 125 IP address on data links 167 IP interfaces on data links 167 local secret M_Key 75 secret M_Key 67 connector autodisabled adding 44 deleting 46 reenabling 45 autodisabling 43 displaying autodisable configured 36 controlling chassis 28 gateway parameters 202 gateway ports 202 I4 switch chip 40 InfiniBand fabric 97 LAGs 183 nodes 101 Subnet Manager 56 VLANs 190 VNICs 190 create_ipoib command 96 135 createfabric add element command 122 delete element command 122 list active command 96 modify element command 123 start command 120 createfabric command 120 createlag command 185 186 createvlan command 192 createvnic command 195 creating central configuration file 154 fabric configuration 127 fabric element configuration 120 Global Gateway 186 IPoIB interface 135 LAGs 185 Oracle Solaris VNICs 169 partition configuration 110 secret M_Key configuration 67 VIOAs 160 163 VLANs 192 VNICs 142 148 195 VNIC specific configuration files 156 D data counters clearing 102 data links 161 desport command 32 delete_ipoib command 136 deletelag command 188 deletevlan command 194 delete
21. By default gateway instance numbers are set according the 6 lowest bits of the gateway IP address and must be unique for VNICs to be operational When two or more gateways share the same InfiniBand fabric they might have duplicate gateway instance numbers In this situation you must assign unique instance numbers to all of the gateways in the fabric 1 On each gateway s management controller display the gateway s instance number FabMan gateway_name gt setgwinstance list 2 Determine which gateway will receive a new instance number and what that number will be The gateway instance is an number from 0 to 63 3 On that gateway s management controller type FabMan gateway_name gt setgwinstance instance where instance is an number from 0 to 63 For example to the set the instance to 3 FabMan gateway_name gt setgwinstance 3 Stopping Bridge Manager OK Starting Bridge Manager OK FabMan gateway_name gt Related Information m Gateway Reference setgwinstance command m Enable a Gateway Ethernet Port on page 203 m Disable a Gateway Ethernet Port on page 204 m Set the Gateway Ethernet Port Parameters on page 205 m Set the Gateway Service Level on page 206 m Set the Gateway System Name on page 208 Administering Gateway Resources 207 a Display the Gateway Instance Number on page 181 Vv Set the Gateway System Name The BridgeX manager within the mana
22. Linux on page 149 m Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 m Determine VLAN Associations for Gateway Manual Mode Linux on page 145 m Configure and Create VNICs for Gateway Manual Mode Linux on page 146 a VNIC Configuration on Linux Hosts in Gateway Manual Mode on page 147 Administering Gateway Resources 143 V Determine VNIC Configuration Parameters For Gateway Manual Mode Linux MAC addresses GUIDs and connector names or LAGs are assigned to VNICs You can create a repository of this information to assist in creating VNICs Note This procedure creates a gateway MAC address list used to create VNICs in gateway manual mode 1 Determine if you will use unique global MAC addresses or locally administrated MAC addresses 2 Create a list of MAC addresses one per row Note Each MAC address must be unique and not 00 00 00 00 00 00 Note Only even numbers are supported for the most significant byte of the MAC address unicast 3 To the left of each MAC address provide the HCA port GUID to receive that MAC address 4 To the left of each GUID and MAC address pair provide the name of the connector or LAG where the host will be physically connected to the 10GbE network There are eight connector names 0A ETH 1 to OA ETH 4 and 1A ETH 1 to 1A ETH 4 5 Use the list for when you create the VNICs An example entry in the gat
23. a Controlling VLANs and VNICs on page 190 m Controlling Gateway Ports and Parameters on page 202 Vv Display LAG Information The showlag command displays general information about all LAGs or detailed information for a specific LAG 1 On the management console display general information about all LAGs FabMan gateway_name gt showlag Name Iport Members Vnics Gw Port ID BANDWIDTH Distribution LACP Mode DiscoverPkey Type SL bxmtest0 Bridge 0 1 0 1000 2704 Oxffft Global 1 FabMan gateway_name gt Administering Gateway Resources 173 2 Display detailed information about a specific LAG FabMan gateway_name gt showlag lagname where lagname is the identifier of the LAG For example FabMan gateway_name gt showlag test enabled active RC FabMan gateway_name gt 3 If the information displayed is incorrect modify the LAGs See Add or Delete Connectors From a LAG on page 187 Related Information m Gateway Reference showlag command a Controlling LAGs on page 183 m Display the VLANs on page 174 m Display the VNICs on page 176 m Display IO Adapters on page 177 a Display Gateway Ethernet Port Information on page 178 a Display Gateway Port Information on page 179 a Display the Gateway Instance Number on page 181 a Display Information About the Gateway on page 182 V Display the VLANs You can use the showvlan command to id
24. gateway MAC address list m lJagname is the identifier of the LAG m vlan_ID is the VLAN identifier NO or 2 to 4094 This value is column four of the gateway MAC address list 192 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN m p_key is the partition identifier 1 to 7fff or default This value is column five of the gateway MAC address list For example to associate connector 1A ETH 3 with VLAN 3 and default partition key FabMan gateway_name gt createvlan 1A ETH 3 vlan 3 pkey default FabMan gateway_name gt 4 Repeat Step 3 for all VLANs to be associated with connectors Note For the createvlan command given a connector lagname and vlan_id combination there is only one p_key permitted 5 Verify the VLAN configuration FabMan gateway_name gt showvlan Connector LAG VLN PKEY OA ETH 1 0 EELE 0A ETH 2 0 ffff 0A ETH 3 0 FELE 0A ETH 4 0 ffff A ETH 1 3 ECEE A ETH 1 0 ffff A ETH 2 3 ffff A ETH 2 0 FEEF 1A ETH 3 3 ffff 1A ETH 3 0 EEEE 1A ETH 4 3 FLEE 1A ETH 4 0 EEEE FabMan gateway_name gt Related Information m Gateway Reference createvlan command m Gateway Reference showvlan command m Delete VLANs on
25. m Disable or Enable an External Port on page 104 Related Information m Investigating Nodes on page 98 m Partitioning the InfiniBand Fabric on page 105 m Correlating Fabric Elements on page 119 m Configuring the Fabric Director Node List on page 125 m Performing Supportive Tasks on page 132 Administering the InfiniBand Fabric 101 Vv Clear Data and Error Counters When you are optimizing the InfiniBand fabric for performance you might want to know how the throughput increases or decreases according to changes you are making Alternatively if you are troubleshooting a port the perfquery command provides counters of errors occurring at that port To determine the throughput as a function of time or if the problem at the port has been resolved the ibdiagnet pc command enables you to reset the data counters and error counters to 0 On the management controller type FabMan gateway_name gt ibdiagnet pc Related Information m Gateway Reference indiagnet command E Gateway Reference perfquery command m Display Counters for a Node on page 89 V Reset a Port You might need to reset a port to determine its functionality Note If secret M_Key functionality is enabled use the M m_key option to the ibportstate command for external ports On the management controller type FabMan gateway_name gt ibportstate M m_key lid port reset where m m_key i
26. on page 93 Related Information m Monitoring the Chassis on page 19 m Monitoring the I4 Switch Chip on page 31 m Monitoring Gateway Resources on page 172 83 Monitoring the Subnet Manager on page 49 Monitoring Gateway Resources on page 172 Controlling the InfiniBand Fabric on page 97 Displaying Information About Nodes These tasks display information about the nodes in the InfiniBand fabric Identify All Switches in the Fabric on page 84 Identify All CAs in the Fabric on page 85 Display the InfiniBand Fabric Topology Simple on page 86 Display the InfiniBand Fabric Topology Detailed on page 86 Display a Route Through the Fabric on page 87 Display the Link Status of a Node on page 88 Display Counters for a Node on page 89 Display Low Level Detailed Information About a Node on page 90 Display Low Level Detailed Information About a Port on page 91 Related Information Investigating Nodes on page 98 Controlling Nodes on page 101 Displaying Fabric Functionality Configuration Information on page 93 V Identify All Switches in the Fabric If you need to know identity information about the switches in the InfiniBand fabric you can use the ibswitches command This command displays the GUID name LID and LMC for each switch The output of the command is a mapping of GUID to LID for switches in the fabric Note
27. the secret M_Key 2 If the information displayed is incorrect modify the parameters See m Set the Subnet Manager Priority on page 56 m Set the Subnet Manager Prefix on page 57 m Enable Subnet Manager Controlled Handover on page 58 m Set the Subnet Manager Readable Management Key on page 59 m Set the Subnet Manager Routing Algorithm on page 61 Related Information m Gateway Reference setsmpriority command m Controlling the Subnet Manager on page 56 m Display Subnet Manager Status on page 50 a Display the Subnet Manager s Location History on page 51 m Verify the Subnet Manager Configuration Integrity on page 54 Display the Subnet Manager Log on page 55 Vv Display the Secret M_Key Configuration You can display the active currently in use or modified pending secret M_Key configuration with the smsubnetprotection list state command Note You must be in a configuration session to display the modified configuration Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 On the management controller that hosts the master Subnet Manager display the secret M_Key configuration FabMan gateway_name gt smsubnetprotection list state where state is either active or modified For example FabMan gateway_name gt smsubnetprotection list modified File _format_version_number 1 Sun DCS IB mkey config
28. vniclimit option sets the maximum number of VNICs to external port associations 1 1024 VNICs 4 4096 VNICs For example FabMan gateway_name gt createlag 0A12 OA ETH 1 OA ETH 2 lag created FabMan gateway_name gt Administering Gateway Resources 185 3 Repeat Step 2 for any other LAGs to be created Related Information m Gateway Reference createlag command m Enable LAG Mode on page 184 m Add or Delete Connectors From a LAG on page 187 m Delete a LAG on page 188 m Disable LAG Mode on page 189 V Create a Global Gateway A Global Gateway is a way to use EoIB functionality internal to the InfiniBand subnet without connections to any specific external port of the gateway Use the createlag command without specifying any connectors to create a Global Gateway Note A Global Gateway can accommodate up to 8000 VNICs and you can create a maximum of 16 Global Gateways per Sun Network QDR InfiniBand Gateway Switch However the Sun Network QDR InfiniBand Gateway Switch only supports a total of 8192 VNICs regardless of how they are distributed to LAGs Global Gateways and external ports 1 Enable LAG mode See Enable LAG Mode on page 184 2 On the management controller create the Global Gateway FabMan gateway_name gt createlag lagname discoverypkey p_key datas1 sl where m lJagname is the identifier of the Global Gateway m p_key is the discovery partition ke
29. 162 Related Information m pkg command man page m Ethernet Over InfiniBand Overview Oracle Solaris on page 161 m Install the Ethernet Over InfiniBand Driver Oracle Solaris on page 162 a Set Up VIOAs Oracle Solaris on page 163 m Configure IP Addresses on the Data Link Oracle Solaris on page 167 m Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 V Install the Ethernet Over InfiniBand Driver Oracle Solaris 1 Consider your first steps 162 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 2 m If you know for certain that your Oracle Solaris 11 image contains the ethernet over ib package go to Step 4 m If you are not certain go to Step 2 Download the Oracle Solaris 11 operating system See this URL for instructions http www oracle com technetwork server storage solaris11 dow nloads index html Install the Oracle Solaris 11 operating system according to the documentation included with the software As superuser of the Oracle Solaris 11 host install the Ethernet over InfiniBand package pkg install ethernet over ib Follow the prompts Related Information pkg command man page Ethernet Over InfiniBand Overview Oracle Solaris on page 161 Verify That the Ethernet Over InfiniBand Driver Is Installed Oracle Solaris on page 162 Set Up VIOAs Oracle Solaris on
30. 9600 Global Global XFI 10Gb s OA ETH 2 Bridge 0 2 Enabled Up Up 9600 Global Global XFI 10Gb s OA ETH 3 Bridge 0 1 Enabled Up Up 9600 Global Global XFI 10Gb s OA ETH 4 Bridge 0 1 Enabled Up Up 9600 Global Global XFI 10Gb s CONNECTOR 1A ETH Port Bridge Adminstate Link State MTU TxPause RxPause Linkmode Speed 180 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 1A ETH 1 Bridge 1 2 Enabled Up Up 9600 Global Global XFI 10Gb s 1A ETH 2 Bridge 1 2 Enabled Up Up 9600 Global Global XFI 10Gb s 1A ETH 3 Bridge 1 1 Enabled Up Up 9600 Global Global XFI 10Gb s 1A ETH 4 Bridge 1 1 Enabled Up Up 9600 Global Global XFI 10Gb s FabMan gateway_name gt When the v option is used this information is provided m BridgeX devices The node GUID and node description are provided m Internal ports The device port port name peer port port GUID LID InfiniBand and gateway state speed and MTU are displayed m Ethernet connectors For each connector the port associated bridge the administrative link and physical state MTU transmit and receive pause link mode and speed are provided Note Your output will differ from that of the example Related Information m Gateway Reference showgwports command m Set the Gateway Ethernet Port Parameters on page 205 a Display LAG Information on page 173 m Display the VLANs on page 174 m
31. Fan 2 running at rpm 11313 Fan 3 running at rpm 11521 Fan 4 not present FabMan gateway_name gt Related Information m Gateway Reference get fanspeed command a Display Gateway Environmental and Operational Data on page 23 Vv Display Gateway Environmental and Operational Data The env_test command enables you to perform several investigative tasks helping you determine the internal environment and operational status of the gateway On the management controller type FabMan gateway_name gt env_test Environment test started Starting Environment Daemon test Environment daemon running Environment Daemon test returned OK Starting Voltage test Voltage ECB OK Measured 3 3V Main 3 28 V Measured 3 3V Standby 3 37 V Measured 12V 12 06 V Measured 5V 5 03 V Measured VBAT 3 25 V Measured 1 0V 1 01 V Measured I4 1 2V 1 22 V Measured 2 5V 2 51 V Measured V1P2 DIG 1 18 V Measured V1P2 ANG 1 18 V Measured 1 2V BridgeX 1 22 V Measured 1 8V 1 80 V Measured 1 2V Standby 1 20 V Voltage test returned OK Starting PSU test Administering the Chassis 23 PSU 0 present PSU 1 present PSU test returned OK Starting Temperature test Back temperature 30 5 Front temperature 30 6 SP temperature 34 1 Switch temperature 48 maxtemperature 49 Bridge 0 temperature 40 maxtemperature 41 Bridge 1 temperature 45 maxtemperature 45 Temperature test returned OK Sta
32. Gateway Resources 175 Vv Display the VNICs The showvnics command displays information about and status of the VNICs 1 On the management controller type FabMan gateway_name gt showvnics ID STATE FLG IOA_GUID NODE IID MAC VLN PKEY GW 16 UP N 00 21 28 00 01 3E CE A0O mnm34 55 0000 02 02 02 02 02 04 3 ffff 1A ETH 4 13 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 01 3 ffff 1A ETH 3 15 UP N 00 21 28 00 01 3E CE 9F mnm34 55 0000 02 02 02 02 02 03 3 ffff 1A ETH 1 14 UP N 00 03 BA 00 01 00 E3 72 mnm34 60 0000 02 02 02 02 02 02 3 ffff 1A ETH 2 FabMan gateway_name gt Note Your output will differ from that in the example For each VNIC the showvnics command displays the VNIC ID its state any flags its associated host GUID and node description its MAC address its associated VLAN ID its partition key and connector 2 If the information displayed is incorrect modify the VNICs See Modify VNICs on page 197 Related Information m Gateway Reference showvnics command a Controlling VLANs and VNICs on page 190 m Display LAG Information on page 173 m Display the VLANs on page 174 m Display IO Adapters on page 177 a Display Gateway Ethernet Port Information on page 178 a Display Gateway Port Information on page 179 a Display the Gateway Instance Number on page 181 a Display Information About the Gateway on page 1
33. Information m Monitoring Gateway Resources on page 172 m Controlling LAGs on page 183 m Controlling VLANs and VNICs on page 190 m Controlling the Chassis on page 28 m Controlling the I4 Switch Chip on page 40 m Controlling the InfiniBand Fabric on page 97 m Controlling the Subnet Manager on page 56 a Installing Gateway Supportive Software Linux on page 137 m Creating VNICs Under Gateway Manual Mode Linux on page 142 m Creating VNICs Under Host Manual Mode Linux on page 148 m Creating Virtual IO Adapters Oracle Solaris on page 160 V Enable a Gateway Ethernet Port On the management controller type FabMan gateway_name gt enablegwport connector discoverpkey p_key where m connector is the name of the connector hosting the port 0A ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH m p_key is the discovery partition key 1 to 7 or default Administering Gateway Resources 203 Note The discoverypkey option ensures that only host ports sharing the discovery partition are able to use this gateway port For example to enable the port associated with connector 0A ETH 3 type FabMan gateway_name gt enablegwport OA ETH 3 FabMan gateway_name gt Related Information a Gateway Reference enablegwport command a Disable a Gateway Ethernet Port on page 204 m Set the Gateway Ethernet Port Parame
34. Linux on page 157 m Configure and Create VNICs for Host Manual Mode Linux on page 158 mlx4_vnic_confd Daemon Linux The m1x4_vnic_confd daemon is used to manage host administrated VNICs The daemon is located in the etc init d directory of each InfiniBand host The syntax for the command line is mlx4_vnic_confd_ operation where operation sets the daemon s interaction with VNICs This table describes the operations supported by the m1x4_vnic_confd daemon Administering Gateway Resources 157 Operation Description start Starts loads or creates new host administrated VNICs stop Stops all host administrated VNICs restart Closes and then reopens all host administrated VNICs reload Updates the system with the most current configuration files This operation does not modify host administrated VNICs with unchanged configuration files Related Information a Host Manual Mode Overview Linux on page 149 m Central Configuration File Linux on page 150 a VNIC Specific Configuration File Linux on page 151 a Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 m Create the Central Configuration File Linux on page 154 m Create the VNIC Specific Configuration Files Linux on page 156 m Configure and Create VNICs for Host Manual Mode Linux on page 158 Vv Configure and Create VNICs for Host Manual Mode Linux 1 On the manageme
35. Monitoring Gateway Resources on page 172 a Controlling LAGs on page 183 m Controlling VLANs and VNICs on page 190 m Controlling Gateway Ports and Parameters on page 202 V Acquire the BXOFED Software Linux 1 Open a web browser on a host that will receive the BXOFED software 2 Go to this URL http support oracle com Oracle s My Oracle Support page is displayed 3 Sign in if you already have an account The dashboard page is displayed Note If you do not have an account you must register 4 Click the Patches amp Updates tab The Patches amp Updates page is displayed 5 In the Patch Search window click the Product or Family Advanced The Patch Search window updates 6 In the Product Is field type BridgeX Possible products are suggested 7 Click the most appropriate link The Release Is field might autopropagate with the most current version 138 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 10 11 12 13 14 15 16 17 In the Release drop down menu select the most current version of the BridgeX OFED software For example BridgeX OFED 1 5 1 Click outside of the drop down menu Click Search The Patch Search window expands with the search results In the Patch Name column click the patch number link respective to your platform For example 12621910 The Patch Sea
36. The driver is delivered through an IPS based package with the file name of ethernet over ib Network interfaces in the Oracle Solaris 11 operating system correspond to instances of network hardware devices and the interfaces are configured over data links The Ethernet over InfiniBand driver supports virtual network hardware on HCAs and enables interfaces to be assigned to that virtual hardware Administrators create IP interfaces on top of data links A data link represents a link object in the second layer of the OSI model The respective physical link is directly associated with a device physical or virtual and device instance name The device instance name is comprised of the device driver name augmented with the instance number which has a value of 0 to n instances of network devices physical or virtual using that driver For each virtual device and interface created on the gateway there is a corresponding data link named eoibX created on the Oracle Solaris 11 host Associating the InfiniBand port of an HCA to a gateway Ethernet port and assigning one or more MAC addresses to the pair creates a virtual IO adapter VIOA The createvnic command of the gateway is used to fulfill this objective Oracle Solaris discovers the VIOA binds an eoibX datalink instance to the VIOA and then manages the VIOA as if it were a physical network interface card NIC Like a physical NIC the VIOA represents the access path to an Ethernet port Because mo
37. The output for your InfiniBand fabric will differ greatly from the output in the example 84 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 On the management controller type FabMan gateway_name gt ibswitches Switch 0x00212856cfe2c0a0 ports 36 SUN IB QDR GW switch mnm34 98 enhanced port 0 lid 6 Imc 0 FabMan gateway_name gt Related Information m Gateway Reference ibswitches command m Gateway Reference ibhosts command m Identify All CAs in the Fabric on page 85 V Identify All CAs in the Fabric Similar to the ibswitches command the ibhosts command displays identity information about the channel adapters CAs inclusive of the host channel adapters HCAs in the InfiniBand fabric The output contains the GUID and name for each CA Note Unlike the ibswitches command the ibhosts command does not display the LIDs 1 On the management controller type Note The output in the example is just a portion of the full output and varies for each InfiniBand topology FabMan gateway_name gt ibhosts Ca 0x00212856cd22c000 ports 2 SUN IB QDR GW switch mnm34 97 Bridge 0 Ca 0x00212856cd22c040 ports 2 SUN IB QDR GW switch mnm34 97 Bridge 1 Ca 0x0002c903000891aa ports 2 mnm34 54 HCA 1 Ca 0x00212800013ece9e ports 2 mnm34 55 HCA 1 Ca 0x0003ba000100eE370 ports 2 mnm34 60 HCA 1 FabMan gateway_name gt 2 Optional
38. VNIC Each VNIC entry has this format and parameters name name mac mac ib_port device port vid vlan_ID wnic_id number bx string eport connector where m name is the VNIC device name or eth number m mac is the MAC assigned to the VNIC m device is the device name retrieved from the hca_id field in the output of the ibv_devinfo command m port is the port number either 1 or 2 a vlan_ID is the VLAN identifier to assign to the VNIC NO or 2 to 4094 Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN Note The vid parameter is optional m number is a unique number to assign to the VNIC The value is 1 to 32767 m string is either the HCA port GUID or the system name m connector is the gateway connector assigned to the VNIC OA ETH 1 to OA ETH 4 and 1A ETH 1 to 1A ETH 4 Related Information a Host Manual Mode Overview Linux on page 149 m VNIC Specific Configuration File Linux on page 151 m Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 m Create the Central Configuration File Linux on page 154 m Create the VNIC Specific Configuration Files Linux on page 156 mw mlx4_vnic_confd Daemon Linux on page 157 m Configure and Create VNICs for Host Manual Mode Linux on page 15
39. about the fabric element through the LDA and ENVD daemons running on the fabric element From the information gathered the Fabric Director creates a local SNMP fabricMIB The Fabric Director of the element hosting the master Subnet Manager is assigned the role of master The master Fabric Director gathers information about all of the fabric elements by communicating out of band management network with the Fabric Director instances on those fabric elements This master Fabric Director creates a model of the fabric that is also in the form of an SNMP MIB as defined by the SUN FABRIC MIB specification By default Fabric Directors run in nonmaster mode Only when a fabric element s Subnet Manager becomes the master Subnet Manager does the fabric element s corresponding Fabric Director become the master Fabric Director This table describes the fabricMIB tables and fabricMIB scalars generated by Fabric Directors Administering the InfiniBand Fabric 125 Type Generated fabricMIB Tables Generated fabricMIB Scalars All Fabric Directors e fabricElemPortTable e fabricElemMgrIpAddress e fabricElemConnectorTable e fabricElemName e fabricElemNumPorts e fabricElemOperStatus e fabricElemNumConnectors Master Fabric Directors e fabricMgmtElemTable e fabricMgmtFabricDescr e fabricMgmtExtElemTable e fabricMgmtFabricType e fabricMgmtGatewayTable e fabricMgmtFabricTopology e fabricMgmtLinkTable e fabricMgmtFabricOperStatus
40. exist in the filesystem of every management controller running a Subnet Manager You create this list with the smodes command The list contains the IP addresses of all active management controllers running a Subnet Manager in your fabric Note If you are using DHCP for your management network you must configure the DHCP server to assign a fixed and unique IP address to the Subnet Manager node based upon the MAC address of that node The algorithm to determine the IP address is user determined It is most important that the IP address is unique and consistent The list should have an entry for every Sun Datacenter InfiniBand Switch 36 and Sun Network QDR InfiniBand Gateway Switch that runs a Subnet Manager in your InfiniBand fabric Note If the Subnet Manager nodes of your InfiniBand fabric ever change disabled added and so on you must update all copies of the Subnet Manager nodes file 1 Determine the IP addresses of all management controllers in your InfiniBand fabric 2 On the management controller hosting the master Subnet Manager propagate the Subnet Manager nodes file with the IP addresses of all Subnet Manager nodes FabMan gateway_name gt smnodes add P_address IP_address where IP_address is the IP address of each management controller hosting a Subnet Manager For example FabMan gateway_name gt smnodes add 123 45 67 89 123 45 67 90 FabMan gateway_name gt 3 Repeat Step 2 for all
41. forward the link 4 Using Switch Chip Port to QSFP Connectors and Link LED Routes on page 8 it is determined that the link routes to connector 15A The LED blinks Related Information m Switch Chip Port to QSFP Connectors and Link LED Routes on page 8 m QSFP Connectors and Link LEDs to Switch Chip Port Routes on page 8 Troubleshooting the Gateway 9 a Display the Switch Chip Port to QSFP Connector Mapping on page 32 a Display a Route Through the Fabric on page 87 63 Switch GUIDs Overview Global unique identifiers GUIDs are unique 64 bit strings that identify nodes such as switches and channel adapters For the Sun Datacenter InfiniBand switches and this gateway GUIDs are modified to identify the node s role and location This table describes the GUID s structure 16 15 12 118 7 4 3 0 MAC 48 bits ype Pos Pevice INum The five fields of the GUID are described as follows m MAC Bits 63 through 16 are the Machine Allocation Code MAC address A standard for network components the MAC address is typically provided by manufacturers in a 6 byte colon delimited string For example 00 11 22 33 44 55 m Type Bits 15 through 12 is the type of board on which the node resides Pos Bits 11 through 8 identify the position of the board within the gateway m Device Bits 7 through 4 identify which device on the board has that node Num Bits 3 through 0 are
42. gateway_name gt smsubnetprotection list modified Smkey Attribute 0x15fc26aea300f831 Ox3fc7e6ad701la8a2a C 8 End the session and commit the secret M_Key configuration to the master Subnet Manager FabMan gateway_name gt FabMan gateway_name gt smsubnetprotection commit The configuration is distributed to the trusted standby Subnet Managers and the current secret M_Key to the InfiniBand Fabric 9 Display the active secret M_Key configuration See Display the Secret M_Key Configuration on page 52 Related Information m Gateway Reference smsubnetprotection command 74 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 m Create the Secret M_Key Configuration on page 67 m Add Secret M_Keys to the Configuration on page 69 m Set the Secret M_Key to Use on page 72 m Configure a Local Secret M_Key on page 75 V Configure a Local Secret M_Key When a gateway or switch from an InfiniBand subnet using secret M_Key functionality is brought into a fabric using a different secret M_Key configuration the gateway is isolated from traffic because the secret M_Keys mismatch In this situation you must provide a recognized secret M_Key to the added gateway so that the gateway is merged into the fabric and secured with the current secret M_key configuration Conversely by manually changing the current secret M_Key of a gateway or swit
43. m Gateway Reference getportstatus command m Display the Link Status of a Node on page 88 Vv Display Switch Chip Port Counters You can check the errors and throughput of a switch chip port using the getportcounters command On the management controller type FabMan gateway_name gt getportcounters port connector where m port is the number of the port 1 to 36 m connector is the name of the connector 0A to 15A OA ETH 1A ETH 0B to 15B 38 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Note The subordinate names of the 1OGbE connectors 0A ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 are also valid connector names For example for port 1 FabMan gateway_name gt getportcounters 1 Port counters for internal Switch port 1 SyMbolErrors sig fee escenario 0 LINKREGCOVELS eed eis sab SHR neces 0 LInkDowne diese ve secede dae ae eee a 0 ROVELY OTS fin ial ees e a AEE E PALS 0 RcevRemotePhysErrors 0 RevSwRelayErrors 0 MMEDPSCAardss lease e a aE TEN bas 0 XmtConstraintErrors 0 RevConstraintErrors 0 LinkIntegrityErrors 0 ExcBufOverrunErrors 0 VLELSDropp d osare Sw as Aeros lee ares 0 MMED SC ais se et a a a g Guan aca aN a 909556 REV DAE aa 2 ii sess Salle seacoast eee ems 737760 MMEPKGS4 ede Hosta tack ets ie da oak 12830 REV PRES H
44. management controllers hosting Subnet Managers in the InfiniBand fabric Administering the InfiniBand Fabric 133 4 Determine how you will partition your InfiniBand fabric See Determine the Partitions and P_Keys on page 109 Related Information m Gateway Reference snmodes command m Display the smnodes List on page 95 m Update the smnodes List on page 134 m Create a User Partition on page 110 m Prepare for Secret M_Key Functionality on page 66 V Update the smnodes List 1 On the management controller of the master Subnet Manager display all the Subnet Manager nodes FabMan gateway_name gt smnodes list 123 45 67 89 123 45 67 90 FabMan gateway_name gt 2 Add any missing Subnet Manager node IP addresses for the local fabric FabMan gateway_name gt smnodes add P_address IP_address where IP_address is the IP address of each management controller hosting a Subnet Manager For example FabMan gateway_name gt smnodes add 123 45 67 91 123 45 67 92 FabMan gateway_name gt 3 Or delete any unnecessary Subnet Manager node IP addresses for the local fabric FabMan gateway_name gt smnodes delete IP_address IP_address where IP_address is the IP address of each management controller hosting a Subnet Manager For example FabMan gateway_name gt smnodes delete 123 45 67 91 123 45 67 92 FabMan gateway_name gt 134 Sun Network QD
45. name HCA GUIDs and port GUIDs The port GUIDs are the HCA GUID 1 port 1 and HCA GUID 2 port 2 respectively 6 Verify that the gateway recognizes the Oracle Solaris 11 host correctly If not check the cabling between the gateway and the Oracle Solaris 11 host 7 Determine if a VIOA has already been assigned to the Oracle Solaris 11 host FabMan gateway_name gt showvnics ID STATE FLG IOA_GUID NODE IID MAC VLN PKEY GW 4 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 03 NO ffff OA ETH 1 5 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0002 02 02 02 02 02 04 NO ffff OA ETH 1 164 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 3 DISABLED N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 02 NO ffff OA ETH 1 6 UP N 00 03 BA 00 01 00 E3 72 mnm34 60 0000 02 02 02 02 02 05 NO ffff OA ETH 1 FabMan gateway_name gt Note Until the IP interface is created on the data link using the ipadm command at the Solaris host the STATE of the VNIC is WAIT IOA 8 Examine the output by column STATE The state of the VIOA IOA_GUID The port GUID of the VIOA The GUID might be listed several times NODE The host name of the node hosting the VIOA MAC The MAC address assigned to the VIOA There might be several MAC addresses PKEY The partition key assigned GW The physical connector cabled to the n
46. numbers reserved for the programs which modify the GUID In most occurrences the value is 0x2 This table provides values for Type Pos Device Board Type Position Device Fabric card OxF 0x0 Fabric card 0 to 0x8 Fabric card 8 OxA I4 chip 0 to 0xB I4 chip 1 Line card 0x1 0x0 Line card 0 to 0x8 Line card 8 OxA I4 chip 0 to 0xD I4 chip 3 Gateway 0xC 0x0 OxA I4 chip 0 0x0 BX chip 0 0x4 BX chip 1 36 Port OxA 0x0 OxA I4 chip 0 72 Port OxB 0x0 OxA I4 chip 0 to OxF I4 chip 5 10 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 For example given this output from the ibnodes command 0x0021283bad45c000 ports 2 SUN IB QDR GW switch gw 2 Bridge 0 The GUID is 0x0021283bad45c000 or 0021283BAD45 C 0 0 0 Using the information provided in this topic a The MAC address is 0x0021283BAD45 or 00 21 28 3B AD 45 m The type is 0xC or a gateway m The position is 0x0 m The device is 0x0 or BridgeX chip 0 m The number is 0 Related Information m Gateway Reference ipnetdiscover command m Gateway Reference ibnodes command m Gateway Reference ipbswitches command m Gateway Reference ibhosts command m Identify All Switches in the Fabric on page 84 m Gateway Hardware Problems on page 1 m InfiniBand Fabric Problems on page 3 m Network Management Troubleshooting Guidelines on page 6 m Understanding Routing Thr
47. or IPoIB For example FabMan gateway_name gt createfabric add element name mnm13 gw 1 ilomIp 123 45 67 90 Element added FabMan gateway_name gt 4 Or delete a fabric element from the configuration FabMan gateway_name gt createfabric delete element name hostname where hostname is the host name of the element For example FabMan gateway_name gt createfabric delete element name mnm13 gw 1 Element deleted FabMan gateway_name gt 5 Display the fabric element configuration FabMan gateway_name gt createfabric list modified Hostname ILOM IP Addr Redundant IP Addr 122 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 mnm12 gw 1 123 45 67 89 E mnm13 gw 1 123 45 67 90 FabMan gateway_name gt 6 End the configuration session and commit the in progress configuration to the active fabric element configuration FabMan gateway_name gt createfabric complete FabMan gateway_name gt 7 Repeat Step 2 through Step 6 on all management controllers hosting Subnet Managers in the InfiniBand fabric Related Information m Gateway Reference createfabric command m Display the Fabric Element Configuration on page 96 m createfabric Command on page 120 m Create a Fabric Element Configuration on page 120 a Modify an Element of the Configuration on page 123 V Modify an Element of the Configu
48. or 2 Note Additional parameters used for regular eth interfaces can be appended to the ifcfg ethXX file Administering Gateway Resources 151 Related Information a Host Manual Mode Overview Linux on page 149 m Central Configuration File Linux on page 150 m Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 m Create the Central Configuration File Linux on page 154 m Create the VNIC Specific Configuration Files Linux on page 156 m mlx4_vnic_confd Daemon Linux on page 157 m Configure and Create VNICs for Host Manual Mode Linux on page 158 V Determine VNIC Configuration Parameters for Host Manual Mode Linux MAC addresses GUIDs and other parameters are assigned to VNICs You can create a repository of this information to assist in creating VNICs Note This procedure creates a host MAC address list used to create VNICs in host manual mode You must associate a VNIC with a VLAN even if that VLAN identifier is NO You also must partition the InfiniBand fabric and provide a partition key for each VLAN If the VLAN identifier NO is to be associated with the P_Key default you need not create the default partition for it already exists Note The association of VLAN identifier NO to partition key default survives firmware upgrades 1 Determine if you will use unique global MAC addresses or locally administrated MAC addre
49. page 126 m Create a Fabric Configuration on page 127 a Modify an Element of the Fabric Configuration on page 131 Y Modify an Element of the Fabric Configuration You can modify the parameters of an element 1 Initiate a fabric configuration session FabMan gateway_name gt fdconfig start fabric config from current FabMan gateway_name gt 2 Modify an element of the fabric configuration FabMan gateway_name gt fdconfig redefine element name name ip IP_address type type role rolel role role2 where m name is the host name of the element as configured by Oracle ILOM m IP_address is the IP address of an element m type is the identifier of the type of element sw36 Sun Datacenter InfiniBand Switch 36 a sw36gw Sun Network QDR InfiniBand Gateway Switch a unknown The element type is unknown m role is the purpose of the element gateway The element acts as a gateway a spineSwitch The element acts as a spine switch ms leafSwitch The element acts as a leaf switch a unknown The element role is unknown For example FabMan gateway_name gt fdconfig redefine element name secondary role spineSwitch Element modified FabMan gateway_name gt Administering the InfiniBand Fabric 131 3 Verify the fabric configuration Name primary secondary FabMan gateway_name gt fdconfig list in progress fabric config IP Addr Type Role s isMaster Fabri
50. page 163 Configure IP Addresses on the Data Link Oracle Solaris on page 167 Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 Set Up VIOAs Oracle Solaris Note If the VIOA is to be assigned to a VLAN you cannot create the VIOA from the gateway Instead use the procedure in Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 1 Become superuser of the Oracle Solaris 11 host For this procedure the Oracle Solaris 11 host name is solaris01 Administering Gateway Resources 163 2 Display the HCA GUIDs and port GUIDs configured on the host dladm show ib LINK HCAGUID PORTGUID PORT STATE PKEYS ibp0 21280001A0A590 21280001A0A591 1 up FFFF ibp1 21280001A0A590 21280001A0A592 2 up FFFF 3 Record the host name HCA GUIDs and port GUIDs In this example the HCA GUID is 21280001A0A590 and the port GUIDs are 21280001A0A591 and 21280001A0A592 4 From the management controller of the gateway display the HCAs recognized by the gateway FabMan gateway_name gt ibhosts Ca 0x0021280001A0A590 ports 2 solaris0O1 Ca 0x00212856cd22c040 ports 2 SUN IB QDR GW switch mnm34 97 Bridge 1 Ca 0x0002c903000891laa ports 2 mnm34 54 HCA 1 Ca 0x00212800013ece9e ports 2 mnm34 55 HCA 1 Ca 0x0003ba000100e370 ports 2 mnm34 60 HCA 1 FabMan gateway_name gt 5 Compare the output of the ibhosts command with the recorded host
51. page 194 m Create VNICs on page 195 m Enable VNICs on page 199 Administering Gateway Resources 193 a Disable VNICs on page 200 m Delete VNICs on page 201 m Display the VLANs on page 174 YV Delete VLANs You can delete VLANs with the deletevlan command 1 Delete all VNICs associated with the connector for the VLAN you are deleting See Delete VNICs on page 201 2 On the management controller type FabMan gateway_name gt deletevian lagname connector vlan vlan_ID where m lJagname is the identifier of the LAG connector is the name of the connector OA ETH 1 to 0A ETH 4 0A 1 1A ETH 1 to 1A ETH 4 and 1A ETH m vlan_ID is the VLAN identifier 2 to 4094 For example to delete the association of connector 1A ETH 1 to VLAN 3 type FabMan gateway_name gt deletevlan 1A ETH 1 vlan 3 FabMan gateway_name gt 3 Repeat from Step 1 for all VLANs to be deleted G1 TH 4 Verify that the VLANs have been deleted FabMan gateway_name gt showvlan Connector LAG VLN PKEY OA ETH 1 0 EEEE 0A ETH 2 0 FEE OA ETH 3 0 FEFE 0A ETH 4 0 Ffff 1A ETH 1 0 Ffff 1A ETH 2 0 EEEE 1A ETH 3 0 Ffff 1A ETH 4 0 Ffff FabMan gateway_name gt 194 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information m Gateway Reference deletevlan command m Gateway Reference showvlan command m Create VLANs
52. page 199 a Disable VNICs on page 200 a Delete VNICs on page 201 m Display the VNICs on page 176 198 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Y Enable VNICs VNICs are automatically enabled when they are created You can use this procedure to enable a previously disabled VNIC 1 Identify the VNIC you want to enable See Display the VNICs on page 176 2 Enable the VNIC on the management controller FabMan gateway_name gt enablevnic connector vnic_ID where connector is the name of the connector OA ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH m vnic_ID is the VNIC identifier For example to enable VNIC 15 on connector 1A ETH 1 type FabMan gateway_name gt enablevnic 1A ETH 1 15 FabMan gateway_name gt 3 Repeat from Step 1 for all VNICs to be enabled 4 Verify that the VNICs have been enabled FabMan gateway_name gt showvnics ID STATE FLG IOA_GUID NODE IID MAC VLN PKEY GW 16 UP N 00 21 28 00 01 3E CE A0 mnm34 55 0000 02 02 02 02 02 04 3 ffff 1A ETH 4 13 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 01 3 ffff 1A ETH 3 15 UP N 00 21 28 00 01 3E CE 9F mnm34 55 0000 02 02 02 02 02 03 3 ffff 1A ETH 1 14 UP N 00 03 BA 00 01 00 E3 72 mnm34 60 0000 02 02 02 02 02 02 3 ffff 1A ETH 2 FabMan gateway_name gt Related Information m Gateway Refe
53. showgwports command smpartition Command on page 106 smpartition Command Guidelines on page 107 Partitions and P_Keys on page 108 Create a User Partition on page 110 Add or Remove a Port From a Partition on page 112 Modify a Partition or Port on page 114 Delete a Partition on page 116 Administering the InfiniBand Fabric 109 m Remove User Partitions for Firmware Downgrade on page 117 Vv Create a User Partition Note After creating user partitions consider modifying the default partition to limited membership for CAs and disabling IPoIB The reason being that nodes with full membership in different user partitions can communicate with each other through the default partition See Modify a Partition or Port on page 114 for an example 1 If you have not already done so create the Subnet Manager nodes list See Create the smnodes List on page 133 2 If the fabric element configuration already exists update the configuration to reflect the smnodes list See Add or Delete an Element From the Configuration on page 122 3 Determine your partition P_Keys See Determine the Partitions and P_Keys on page 109 4 Initiate a partition configuration session on the management controller FabMan gateway_name gt smpartition start FabMan gateway_name gt 5 Create the user partition FabMan gateway_name gt smpartition create n part
54. the configuration 5 Display the fabric element configuration FabMan gateway_name gt createfabric list modified Hostname ILOM IP Addr Redundant IP Addr mnm12 gw 1 123 45 67 89 FabMan gateway_name gt 6 End the configuration session and commit the in progress configuration to the active fabric element configuration FabMan gateway_name gt createfabric complete FabMan gateway_name gt 7 Repeat Step 2 through Step 6 on all management controllers hosting Subnet Managers in the InfiniBand fabric Related Information m Gateway Reference createfabric command m Display the Fabric Element Configuration on page 96 m createfabric Command on page 120 m Add or Delete an Element From the Configuration on page 122 a Modify an Element of the Configuration on page 123 Administering the InfiniBand Fabric 121 V Add or Delete an Element From the Configuration 1 Verify the smnodes list See Display the smnodes List on page 95 2 Initiate a fabric element configuration session on the management controller FabMan gateway_name gt createfabric start from current FabMan gateway_name gt 3 Add a fabric element to the configuration FabMan gateway_name gt createfabric add element name hostname ilomIp IP_address redundantIP P_address where m hostname is the host name of the element a IP_address is the IP address for the element recognized by Oracle ILOM
55. 0 FabMan gateway_name gt 2 If the information displayed is incorrect modify the smnodes configuration list See Update the smnodes List on page 134 Related Information m Gateway Reference smmodes command m Create the smnodes List on page 133 m Update the smnodes List on page 134 Administering the InfiniBand Fabric 95 Vv Display the Fabric Element Configuration 1 On the management controller type FabMan gateway_name gt createfabric list active Hostname ILOM IP Addr Redundant IP Addr mnm12 gw 1 123 45 67 89 mnm13 gw 1 123 45 67 91 FabMan gateway_name gt 2 If the information displayed is incorrect modify the fabric element configuration See Modify an Element of the Configuration on page 123 Related Information m Gateway Reference createfabric command m Create a Fabric Element Configuration on page 120 Vv Display the IPoIB Interface 1 On the management controller type FabMan gateway_name gt create_ipoib s Name PKey IP address Subnetmask Interface Interface type None 0x7fff 123 45 67 89 255 255 255 0 ib0 external FabMan gateway_name gt 2 If the information displayed is incorrect modify the IPoIB interfaces See m Delete the IPoIB Interface on page 136 m Create the IPoIB Interface on page 135 Related Information m Gateway Reference create_ipoib command m Delete the IPoIB Interface on page 136 m Crea
56. 0 01 3E CE 9F mnm34 55 0000 02 02 02 02 02 88 3 ffff 1A ETH 1 14 UP N 00 03 BA 00 01 00 E3 72 mnm34 60 0000 02 02 02 02 02 02 3 ffff 1A ETH 2 FabMan gateway_name gt Related Information m Gateway Reference disablevnic command m Gateway Reference showvnics command a VNIC Allocation on the Gateway on page 191 m Create VLANs on page 192 m Delete VLANs on page 194 m Create VNICs on page 195 m Enable VNICs on page 199 m Delete VNICs on page 201 m Display the VNICs on page 176 V Delete VNICs Before you can delete a VLAN you must delete the VNIC associated with that VLAN and connector 1 Identify the VNIC you want to delete See Display the VNICs on page 176 2 Delete the VNIC on the management controller FabMan gateway_name gt deletevnic connector vnic_ID where m connector is the name of the connector 0A ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH m vnic_ID is the VNIC identifier For example to delete VNIC 15 on connector 1A ETH 1 type FabMan gateway_name gt deletevnic 1A ETH 1 15 FabMan gateway_name gt 3 Repeat from Step 1 for all VNICs to be deleted Administering Gateway Resources 201 4 Verify that the VNICs have been deleted FabMan gateway_name gt showvnics ID STATE FLG IOA_GUID NODE IID MAC VLN PKEY GW 16 UP N 00 21 28 00 01 3E CE A0O mnm34 55 0000 02 02 02 02 02 04 3 fff
57. 02 Partitioning the InfiniBand Fabric These topics enable you to partition the InfiniBand fabric and modify the partition s configuration smpartition Command on page 106 smpartition Command Guidelines on page 107 Partitions and P_Keys on page 108 Determine the Partitions and P_Keys on page 109 Create a User Partition on page 110 Add or Remove a Port From a Partition on page 112 Modify a Partition or Port on page 114 Administering the InfiniBand Fabric m Delete a Partition on page 116 m Remove User Partitions for Firmware Downgrade on page 117 Related Information m Investigating Nodes on page 98 m Controlling Nodes on page 101 m Correlating Fabric Elements on page 119 m Configuring the Fabric Director Node List on page 125 m Performing Supportive Tasks on page 132 smpartition Command You can partition your InfiniBand fabric using the smpartition command and its subcommands The smpartition command is issued from the gateway or switch that has the master Subnet Manager and acts upon the partition configuration There are two configurations the active configuration is the one currently in use and the modified configuration is the one that you can affect You begin a configuration session with the smpartition start command You then make changes to the modified configuration with the smpartition create smpartition add smpartition re
58. 1 Sun DCS IB mkey config file This file is generated do not edit secretmkey enabled nodeid o4nm2 gw 6 time 15 Sep 03 38 33 checksum 0 commit_number 2 He F HH HE Mkey Untrusted Mkey Smkey Attribute 0x00abcdefabcdef01 Oxlaa45124fee612ae Ox15fc26aea300f831 0x00abcdefabcdef02 0x4ccd8230de6cd348 Ox3fc7e6ad701la8a2a S 0x00abcdefabcdef03 Ox9baaldebcc74de5e 0x1b253003600d137b C FabMan gateway_name gt 72 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Note The S in the Attribute column means that M_Key is in standby state and is to become the current secret M_Key 6 End the session and commit the secret M_Key configuration to the master Subnet Manager FabMan gateway_name gt smsubnetprotection commit FabMan gateway_name gt The configuration is distributed to the trusted standby Subnet Managers and the current secret M_Key to the InfiniBand Fabric Related Information m Gateway Reference smsubnetprotection command m Create the Secret M_Key Configuration on page 67 m Add Secret M_Keys to the Configuration on page 69 m Delete Secret M_Keys From the Configuration on page 73 m Configure a Local Secret M_Key on page 75 V Delete Secret M_Keys From the Configuration Note If you delete the secret M_Key also used by a node temporarily unavailable to the fabric when that node becomes available there is a secret
59. 104 Temporary solution e If the problem still remains disable the affected port See Disable or Enable an External Port on page 104 Permanent solution e If the problem still remains replace the affected component or the gateway Refer to Gateway Service servicing data cables Refer to remote port s documentation for replacement procedures Refer to Gateway Service removing the gateway from the rack Refer to Gateway Installation installing the gateway into the rack 1 Disconnect and properly reconnect both ends of the respective InfiniBand cable Refer to Gateway Service servicing the data cables 2 If the LED is still blinking determine the significance of the errors through use of the ibdiagnet command See Determine Which Links Are Experiencing Significant Errors on page 100 3 Determine which connectors map to the affected link by deconstructing the node s GUID and port See Locate a Switch Chip or Connector From the GUID and Port on page 33 4 If some of the links are running at 1x or SDR use that situation elsewhere in this table to rectify the problem 5 Disable and re enable the respective ports See Disable or Enable an External Port on page 104 6 If the errors are still significant swap the cable with a known good one or reconnect the cable to a known good remote port and repeat from Step 2 7 Depending upon what does or does not rectify the problem replace that component
60. 26 Vv Display Power Supply FRU ID The showpsufru command displays power supply FRU ID information Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 On the management controller type FabMan gateway_name gt showpsufru slot where slot is the power supply slot 0 or 1 For example FabMan gateway_name gt showpsufru 0 Sun_SpecPartNo 885 1165 02 UNIX_Timestamp32 Sun Jan 3 15 35 39 2010 Sun_Part_Number 3002143 Sun_Serial_No BFISWA Vendor_ID_Code O2a2 Initial_HW_Dash_Level 02 PSU_Voltage 1 0x04b0 1200 PSU_Current 1 Ox186a 6250 PSU_Voltage 2 0x014a 330 PSU_Current 2 Ox012c 300 Sun_Hazard_Class_Code s Y IPMI_Board_Manufacturer EMERSON IPMI_Board_Product_Name A237 IPMI_Board_Serial_Number 1357ZHO 0952BF15WA IPMI_Board_Part_Number 300 2143 02 FabMan gateway_name gt Related Information m Gateway Reference showpsufru command m Display Chassis FRU ID on page 26 Vv Display Gateway Firmware Versions In time newer or updated gateway firmware might become available The version command displays the versions of the firmware within the gateway chassis On the management controller type FabMan gateway_name gt version SUN DCS gw version 2 1 2 1 Build time Dec 7 2012 09 46 50 FPGA version 0x34 SP board info Manufacturing Date 2009 06 22 Serial Number NCD3R0391 Hardware Revisio
61. 40 m Enable a Switch Chip Port on page 41 Autodisable Functionality Switch chip ports and their connectors can be configured to automatically disable should their links exhibit high error rates or sub optimal link speed or width You use the autodisable command to add the connectors to the autodisable list which has two parts one for connectors whose links fail from high error rates and another for connectors whose links fail from suboptimal link speed or width A connector can be configured for both parts The autodisable feature monitors the following to determine if a connector and its respective link are experiencing high error rates m SNMP traps m Oracle ILOM event log m Syslog m Email alerts The autodisable feature also monitors the link speed and width and if any of the following combinations are discovered the link is considered suboptimal m 1xSDR m 1x DDR 1x QDR m 4x SDR m 4x DDR Administering the 14 Switch Chip 43 Note If you intentionally configure a link for less than 4x QDR do not add that connector to the link speed and width part of the autodisable list It is automatically disabled Autodisable functionality is not enabled by default you must actively configure it by adding connectors to the autodisable list Additionally you must actively monitor its operation and take corrective action if necessary Failure to respond to autodisabled connectors and their links results in a degraded I
62. 5A 15B FabMan gateway_name gt Related Information m Gateway Reference autodisable command m Autodisabling Ports and Connectors on page 43 Vv Display Switch Chip Port Status If by using other procedures it seems that a particular port of the switch chip is problematic you can use the getportstatus command to provide the port state width and speed On the management controller type FabMan gateway_name gt getportstatus connector ibdev port where m connector is the name of the connector 0A to 15A 0A ETH 1A ETH 0B to 15B Note The subordinate names of the 10GbE connectors 0A ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 are also valid connector names m ibdev is the InfiniBand device name Switch Bridge 0 0 Bridge 0 1 Bridge 1 0 Bridge 1 1 m portis the number of the port 1 to 36 For example FabMan gateway_name gt getportstatus Switch 7 Port status for connector 14A Switch Port 7 Enabled Administering the 14 Switch Chip 37 LinkSpeedSupported 66 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LINKS Cate S ca een aiai ere i a ene ewe ek Active PhyYSLiInkStates cies cose ace aiaa paa wi LinkUp LinkSpeedActive 0004 10 0 Gbps LinkSpeedEnabled 2 5 Gbps or 5 0 Gbps or 10 0 Gbps FabMan gateway_name gt Note The parameters with the string Active indicate the current conditions Related Information
63. 67 90 type sw36 role leafSwitch Element added FabMan gateway_name gt 3 Or remove an element from the fabric configuration FabMan gateway_name gt fdconfig remove element name name where name is the identifier of the element For example FabMan gateway_name gt f dconfig remove element name secondary Element deleted FabMan gateway_name gt 4 Verify the fabric configuration FabMan gateway_name gt fdconfig list in progress fabric config Name IP Addr Type Role s isMaster Fabricname primary 123 45 67 89 sw36gw gateway test secondary 123 45 67 90 sw36 leafSwitch no test FabMan gateway_name gt 5 End the configuration session and commit to the new configuration FabMan gateway_name gt fdconfig complete fabric config FabMan gateway_name gt 6 Perform Step 1 through Step 5 on the management controllers of all elements of the fabric 7 If you added a new element create a fabric configuration on the management controller of that element using the same configuration information See Create a Fabric Configuration on page 127 Related Information m Gateway Reference dconfig command a Display the InfiniBand Fabric Configuration on page 94 a Fabric Director and Fabric Elements on page 125 130 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m fdconfig Command Overview on
64. 8 150 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 VNIC Specific Configuration File Linux The ifcfg ethXX file already exists as a means for the network service to derive information about a node The XX represents the eth number of the VNIC To configure VNICs additional parameters are added to the base file format For Red Hat Linux the ifcfg ethXX file has this format DEVICE name HWADDR mac BOOTPROTO dhcp ONBOOT yes BXADDR string BXEPORT connector VNICVLAN vlan_ID VNICIBPORT device port where m name is the VNIC device name or eth number Note The DEVICE parameter is optional If the parameter is missing the suffix of the configuration file name ethXX is used m mac is the MAC assigned to the VNIC m string is either the HCA port GUID or the system name m connector is the gateway connector assigned to the VNIC OA ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 m vlan_ID is the VLAN identifier to assign to the VNIC NO or 2 to 4094 Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN Note The VNICVLAN parameter is optional m device is the device name retrieved from the hca_id field in the output of the ibv_devinfo command m port is the port number either 1
65. 82 176 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Vv Display IO Adapters The showioadapters command displays information about the IO adapters IO adapters represent an HCA port GUID where an EoIB driver is active on the host side 1 On the management controller type FabMan gateway_name gt showioadapters IOA_GUID NODE LID FLAGS GW 0021280001CF205B o4test56 1 HD 1A ETH 2 0021280001CF3787 o4test57 2 HD 1A ETH 2 0021280001CF205B o4test56 1 HD 1A ETH 1 0021280001CF3787 o4test57 2 HD OA ETH 3 0021280001CF205B o4test56 1 HD OA ETH 4 0021280001CF3787 o4test57 2 HD OA ETH 4 FabMan gateway_name gt Note Your output will differ from that in the example The FLAGS indicate whether the VNIC was created by Host Manual Mode HD Gateway Manual Mode ND or other means no flag If the information displayed is incorrect modify the VNICs See Modify VNICs on page 197 or Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 Related Information Gateway Reference showioadapters command Controlling VLANs and VNICs on page 190 Display LAG Information on page 173 Display the VLANs on page 174 Display the VNICs on page 176 Display Gateway Ethernet Port Information on page 178 Display Gateway Port Information on page 179 Display the Gateway Instance Number on page 181
66. ADDR string BXEPORT connector VNICVLAN vlan_ID VNICIBPORT device port where m name is ethXX and XX is column one of the host MAC address file m mac is column two of the host MAC address list m string is column six of the host MAC address list m connector is column eight of the host MAC address list a vlan_ID is column seven of the host MAC address list Note The VNICVLAN parameter is optional m device is column four of the host MAC address list 156 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m port is column five of the host MAC address list For example DEVICE eth1 HWADDR 00 30 48 7d de e4 BOOTPROTO dhcp ONBOOT yes BXADDR 0003ba000100c70b BXEPORT 0A ETH 1 VNICVLAN 0 VNICIBPORT m1x4_0 1 5 Save the file with the name of ifcfg ethXX where XX is the value of column one For example ifcfg ethl 6 Repeat Step 3 to Step 5 for each entry appropriate for the host 7 Repeat Step 2 to Step 6 for each host to have VNIC configuration files Related Information m Create the Central Configuration File Linux on page 154 m Host Manual Mode Overview Linux on page 149 m Central Configuration File Linux on page 150 m VNIC Specific Configuration File Linux on page 151 m Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 m mlx4_vnic_confd Daemon
67. B driver 162 instance number displaying 181 setting 207 investigating nodes 98 ipadm command 167 IPoIB interface creating 135 deleting 136 displaying 96 L LAGs 173 adding connectors 187 controlling 183 creating 185 deleting 188 deleting connectors 187 disabling 189 displaying 173 enabling 184 limited membership 109 link status 35 status of node 88 listlinkup command 35 localmkeypersistence command 53 76 M management controller restarting 29 mapping GUID to CXP connector 33 to switch chip 33 LED to switch chip port 8 QSFP to switch chip port 8 32 switch chip port to LED 8 to QSFP 8 32 matchtopology command 99 membership full 109 limited 109 mlx4_vnic_confd daemon 157 mode gateway manual 142 host manual 148 modifying element 131 fabric elements of configuration 123 partition configuration 114 port membership 114 VNICs 197 modifyvnic command 197 monitoring chassis 19 gateway resources 172 14 switch chip 31 InfiniBand fabric 83 Subnet Manager 49 N NET MGT port troubleshooting 6 nodes controlling 101 counters 89 displaying information 84 information 90 investigating 98 link status 88 O Oracle ILOM command overview 16 Oracle Solaris VNICs 169 overview administrative commands 13 autodisabled connectors 43 Ethernet over InfiniBand 161 gateway manual mode 143 hardware commands 16 host manual mode 149 InfiniBand
68. Configuration on page 123 Related Information m Investigating Nodes on page 98 m Controlling Nodes on page 101 m Partitioning the InfiniBand Fabric on page 105 m Configuring the Fabric Director Node List on page 125 m Performing Supportive Tasks on page 132 Administering the InfiniBand Fabric 119 120 createfabric Command The fabric element configuration file enables both the partition daemon and the Fabric Director to communicate via Ethernet or IPoIB in a redundant fashion This is accomplished by mapping a correlation of the management controller s host name with the IP address recognized by Oracle ILOM and the IPoIB IP address of each element You can use the createfabric command and its subcommands to create or edit the fabric configuration file add modify or delete fabric elements from the file and complete the configuration file and commit it to use You can also display the fabric element configuration file with the createfabric command The createfabric command is issued on each management controller that hosts a Subnet Manager Like the Subnet Manager nodes list the fabric element configuration file that is created on one management controller must be duplicated on all other management controllers There are two fabric element configurations the active configuration is the one currently in use and the in progress configuration is the one that you can affect with the createfabric
69. Display Information About the Gateway on page 182 Administering Gateway Resources 177 Vv Display Gateway Ethernet Port Information You can display Ethernet port information with the setgwethport command The output is similar to that of the getportstatus command 1 On the management controller type FabMan gateway_name gt setgwethport connector where connector is either OA ETH or 1A ETH For example Note Your output will differ from that of the example FabMan gateway_name gt setgwethport OA ETH Port status for connector OA ETH 1 Adminstate Enabled SCALE sss 4 as KES aS ee aw EROS Re SM AS Up Link SACS sic edd aw eel Pea ee Up PrOLOCOL sock eee ewe edutewa dasa eds BEhernet GENK MOGs sieri aie a a e e a et Saas XFI Speedo ks hs ot ae Ee ee od ee hea 10Gb s MIU ened dates So oeb meee EEE es 9600 Te pause ea ee Global RX pasen eee eb tited ve wach dees ae Global Port status for connector OA ETH 2 Adminstate Enabled MDW eea giad aa ate hig mee RN ee Boe see Ae 9600 TX DAUS Cis o Sie dns ode Se ev es ged Global RE Pauses cde a SR bbe ee Ae eR ele Global FabMan gateway_name gt 2 If the information displayed is incorrect modify the gateway Ethernet port parameters See Set the Gateway Ethernet Port Parameters on page 205 Related Information m Gateway Reference setgwethport command a Gateway Reference getpor
70. ILOM CLI m Set the Subnet Manager Priority on page 56 m Set the Subnet Manager Prefix on page 57 m Enable Subnet Manager Controlled Handover on page 58 m Set the Subnet Manager Readable Management Key on page 59 m Clear the Subnet Manager Readable Management Key on page 60 m Set the Subnet Manager Routing Algorithm on page 61 a Enable the Subnet Manager on page 62 a Disable the Subnet Manager on page 63 Related Information m Controlling the Chassis on page 28 m Controlling the I4 Switch Chip on page 40 a Controlling Gateway Ports and Parameters on page 202 m Controlling the InfiniBand Fabric on page 97 m Monitoring the Subnet Manager on page 49 m Controlling the Subnet Manager on page 56 V Set the Subnet Manager Priority By default the Subnet Manager within the management controller is set to 0 priority If there is more than one Subnet Manager in your InfiniBand fabric you must set the priority of each Subnet Manager appropriately The Subnet Manager with the highest priority is the primary or Master Subnet Manager 56 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 1 2 On the management controller disable the Subnet Manager See Disable the Subnet Manager on page 63 Set the Subnet Manager priority FabMan gateway_name gt setsmpriority priority where prio
71. IOA is active FabMan gateway_name gt showvnics ID STATE FLG IOA_GUID NODE IID MAC VLN PKEY GW 7 WAIT IOA N 00 21 28 00 01 A0 A5 91 solaris0O1 0000 a0 a5 91 95 30 9a NO ffff OA ETH 2 4 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 03 NO ffff OA ETH 1 5 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0002 02 02 02 02 02 04 NO ffff OA ETH 1 3 DISABLED N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 02 NO ffff OA ETH 1 6 UP N 00 03 BA 00 01 00 E3 72 mnm34 60 0000 02 02 02 02 02 05 NO ffff OA ETH 1 FabMan gateway_name gt Note Until the IP interface is created on the data link using the ipadm command at the Solaris host the STATE of the VNIC is WAIT IOA 13 Configure an IP address on the data link See Configure IP Addresses on the Data Link Oracle Solaris on page 167 166 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information m dladm command man page m Gateway Reference ibhosts command m Gateway Reference showvnics command m Gateway Reference createvnics command m Ethernet Over InfiniBand Overview Oracle Solaris on page 161 m Verify That the Ethernet Over InfiniBand Driver Is Installed Oracle Solaris on page 162 m Install the Ethernet Over InfiniBand Driver Oracle Solaris on page 162 m Configure IP Addresses on the Data Link Oracle Solaris on page 167 m Creat
72. MTU m pause is the priority flow control pause in quanta 0 to 255 Global or None For example to set the MTU of connector 1A ETH to 9600 type FabMan gateway_name gt setgwethport 1A ETH mtu 9600 Port status for connector 1A ETH 1 AGMINSTACS ee cscri ele wae cee eee ee Enabled SEBS ye Mie eee ee ee ee eee Up Link Statessa ia eia iaa REE Up PrOLOCOh i sreg enere ekaa enh aE Ethernet Link MOd6 toed peace eh ea ee es XFI Speeds peoa a a l ice Ba bree eh aaa ENTE 10Gb s DOP iG og Soe Se beg aoe S Seek de ended See 9600 TH DAUSEs oss 4c fark we BS oes Global RX PAUSE ie 5 oes Ahee ce bi en a Se he Sicee Sees Global Port status for connector 1A ETH 2 Port status for connector 1A ETH 4 Adminstate eies aaeei bee et es Enabled DUALS t 6 kw ie we ow ee Oe ee eS Up Link Statene dent a eea a EA EEN Up PEOLOCO sweets e ia ga a ees A Ethernet Hink MOAG se prestasi ee eRe baa ee XFI SPSS Sco daea a be eee oo ae wien eae 10Gb s MP so acs eek alice Suge ere ave Sable egg eda hte se 9600 T DAUS Gi doe er a Suet cides eke dow ete etn Global RX Daere eas a wee piede lene enh act ee Global FabMan gateway_name gt Administering Gateway Resources 205 Related Information m Gateway Reference setgwethport command a Enable a Gateway Ethernet Port on page 203 a Disable a Gateway Ethernet Port on page 204 m Set the Gateway Service Level on page 206 m Set the Gateway Instance Number on page 207
73. M_Key mismatch and the node is isolated See Configure a Local Secret M_Key on page 75 1 Optional If you want to delete the current secret M_Key in use set the new current M_Key to one you will not delete See Set the Secret M_Key to Use on page 72 2 Display the active secret M_Key configuration See Display the Secret M_Key Configuration on page 52 3 Determine which M_Key you will delete 4 Initiate a secret M_Key configuration session on the management controller that hosts the master Subnet Manager FabMan gateway_name gt smsubnetprotection start deleteonly FabMan gateway_name gt Administering the Subnet Manager 73 5 Delete the M_Key from the configuration FabMan gateway_name gt smsubnetprotection delete m_key where m_key is the management key 16 hexadecimal digits For example FabMan gateway_name gt smsubnetprotection delete 0x00abcdefabcdef03 Element deleted FabMan gateway_name gt 6 Repeat Step 5 for any additional M_Keys you want to delete from the configuration 7 Display the pending M_Keys File _format_version_number 1 Sun DCS IB mkey config file This file is generated do not edit secretmkey enabled nodeid o4nm2 gw 6 time 15 Sep 04 18 20 checksum 0 commit_number 5 He 4b 3 OH OH Mkey Untrusted Mkey 0x00abcdefabcdef01 Oxlaa45124fee612ae 0x00abcdefabcdef02 0x4ccd8230de6cd348 FabMan gateway_name gt FabMan
74. Man gateway_name gt ibswitches Switch 0x03ba01234567a0a0 ports 36 SUN DCS 36P QDR mnm 36p 2 10 172 144 66 enhanced port 0 lid 1 Imc 0 Switch 0x002128312345c0a0 ports 36 SUN IB QDR GW switch mnm gw 1 10 172 144 69 enhanced port 0 lid 15 Imc 0 FabMan gateway_name gt 3 If the GUID is that of a switch truncate all digits of the GUID except the four digits on the right For example GUID 0x002128312345c0a0 is truncated to c0a0 From Switch GUIDs Overview on page 10 the c0 means that the node is inside of a Sun Network QDR InfiniBand Gateway Switch The a means that the node is located in I4 switch chip A the only switch chip The port is 21 4 Use the tables in Understanding Routing Through the Gateway on page 7 to cross reference the routing between ports and to connections For example using Switch Chip Port to QSFP Connectors and Link LED Routes on page 8 I4 switch chip port 21 routes to connector 1B Related Information a Switch GUIDs Overview on page 10 m Gateway Reference dcsport command V Display Switch Chip Boot Status If you believe there is a problem with the switch chip the checkboot command provides a quick pass fail check of the switch chip Should the switch chip fail the test you can use the ibdevreset command in an attempt to reboot the switch chip 34 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 On the ma
75. Oxabbababe M_Key Oxabbababe0001 Routing engine minhop FabMan gateway_name gt 3 Enable the Subnet Manager See Enable the Subnet Manager on page 62 Related Information m Gateway Reference setsmrouting command m Set the Subnet Manager Priority on page 56 m Set the Subnet Manager Prefix on page 57 a Enable Subnet Manager Controlled Handover on page 58 m Set the Subnet Manager Readable Management Key on page 59 m Clear the Subnet Manager Readable Management Key on page 60 a Enable the Subnet Manager on page 62 a Disable the Subnet Manager on page 63 Vv Enable the Subnet Manager The Subnet Manager within the management controller is not enabled by default Use the enablesm command to enable and start the Subnet Manager On the management controller type FabMan gateway_name gt enablesm Starting IB Subnet Manager OK Starting partitiond daemon OK FabMan gateway_name gt 62 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information m Gateway Reference enablesm command m Disable the Subnet Manager on page 63 m Set the Subnet Manager Priority on page 56 m Set the Subnet Manager Prefix on page 57 m Enable Subnet Manager Controlled Handover on page 58 m Set the Subnet Manager Readable Management Key on page 59 m Clear the Subnet Manager Readable M
76. R InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 4 If you are combining fabrics repeat Step 2 adding all the Subnet Manager node IP addresses from the remote fabrics 5 Verify the updated list FabMan gateway_name gt smnodes list 123 45 67 89 123 45 67 90 123 45 67 91 123 45 67 92 123 45 66 77 123 45 68 10 FabMan gateway_name gt 6 Repeat Step 1 through Step 5 for all management controllers hosting Subnet Manager nodes in the local fabric 7 If you are combining fabrics repeat Step 1 through Step 5 for all management controllers hosting Subnet Manager nodes in the remote fabrics Related Information m Gateway Reference smodes command m Display the smnodes List on page 95 m Create the smnodes List on page 133 m Create a User Partition on page 110 m Prepare for Secret M_Key Functionality on page 66 V Create the IPoIB Interface This procedure creates an Internet protocol over InfiniBand interface Note Only one IPoIB interface can be defined On the management controller type FabMan gateway_name gt create_ipoib n if_name m mask t if_type p_key IP_nddress where m if_name is the identifier of the IP interface m mask is the subnet mask m if_type is the type of interface external internal or both Administering the InfiniBand Fabric 135 m p_key is the partition key 1 to 7 f or default m IP_addres
77. Refer to Gateway Service servicing the data cables Refer to remote port s documentation for replacement procedures 4 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Situation Corrective Steps Some InfiniBand links are running at 1x or SDR There are errors on some InfiniBand links For a temporary solution 1 Identify the suspect links using the ibdiagnet command See Find 1x SDR or DDR Links in the Fabric on page 100 Look for text like this W link with SPD 2 5 found at direct path 1 19 From a Switch PortGUID 0x00066a00d80001dd Port 19 To a Switch PortGUID 0x00066a00d80001dd Port 24 2 Determine which connectors map to the affected link by deconstructing the node s GUID and port See Locate a Switch Chip or Connector From the GUID and Port on page 33 3 Verify the cable connection at both ends Refer to Gateway Service servicing the data cables 4 Disable and re enable the respective ports See Disable or Enable an External Port on page 104 5 If the previous steps do not rectify the problem disable the port See Disable or Enable an External Port on page 104 For a permanent solution 1 Perform the steps for a temporary solution Steps 1 to Step 4 2 Swap the cable with a known good cable or reconnect the cable to a known good remote port and repeat from Step 1 3 Depending upon what does or does not rectify the pr
78. Remove all entries from the Subnet Manager node list FabMan gateway_name gt smnodes delete P_address IP_address where IP_address are the IP addresses of the Subnet Manager nodes For example FabMan gateway_name gt smnodes delete 123 45 67 89 123 45 67 90 FabMan gateway_name gt 7 Repeat Step 6 on the management controllers of all Subnet Managers in the InfiniBand fabric 8 Downgrade the firmware Refer to Gateway Remote Management upgrading the firmware 118 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information m Gateway Reference smpartition command m Gateway Reference smmodes command a smpartition Command on page 106 a smpartition Command Guidelines on page 107 m Partitions and P_Keys on page 108 m Determine the Partitions and P_Keys on page 109 m Create a User Partition on page 110 m Add or Remove a Port From a Partition on page 112 m Modify a Partition or Port on page 114 m Delete a Partition on page 116 m Delete VNICs on page 201 m Delete VLANs on page 194 Correlating Fabric Elements These topics help you to create a fabric element configuration m createfabric Command on page 120 m Create a Fabric Element Configuration on page 120 m Add or Delete an Element From the Configuration on page 122 m Modify an Element of the
79. Save the output of ibhosts command as a text file named hostGUIDs txt Administering the InfiniBand Fabric 85 Related Information m Gateway Reference ibhosts command m Gateway Reference ibswitches command m Identify All Switches in the Fabric on page 84 Vv Display the InfiniBand Fabric Topology Simple To understand the routing that happens within your InfiniBand fabric the showtopology command displays the node to node connectivity The output of the command depends on the size of your fabric On the management controller type FabMan gateway_name gt showtopology SUNIBQDRGW mnm34 98 Bridge 0 BX1 P1 4x 10G gt I4_GENERIC mnm34 98 P4 4x 10G gt I4 GENERIC mnm34 98 P3 BX3 P1 4x 10G gt I4_GENERIC mnm34 98 P2 4x 10G gt I4_GENERIC mnm34 98 P1 DEV26428_02P mnm34 55 HCA 1 P1 4x 10G gt SUNIBQDRGW mnm34 97 C 7A P2 4x 10G gt I4_ GENERIC mnm34 98 P33 SUNIBQODRGW mnm34 97 A SW P2 4x 10G gt SUNIBQDRGW mnm34 97 BX3 P1 A SW P4 4x 10G gt SUNIBQDRGW mnm34 97 BX1 P1 C 0B 4x 10G gt I4 GENERIC mnm34 98 P21 FabMan gateway_name gt Related Information m Gateway Reference showtopology command m Determine Changes to the InfiniBand Fabric Topology on page 99 m Perform Comprehensive Diagnostics for the Entire Fabric on page 98 Vv Display the InfiniBand Fabric Topology Detailed To understand the routing that happens within your InfiniBand fabric the ibnetdiscover
80. Solaris Note Follow this procedure to configure Ethernet over InfiniBand with VLAN tagging in Oracle Solaris This procedure is independent of and unrelated to Host Manual Mode or Gateway Manual Mode To create VIOAs on the gateway see Set Up VIOAs Oracle Solaris on page 163 1 Create the VIOAs to use as IP interfaces for the Solaris VNICs See Set Up VIOAs Oracle Solaris on page 163 2 On the management controller create a VLAN to P_Key mapping for the VIOAs FabMan gateway_name gt createvlan connector vlan vlan_ID pkey p_key where Administering Gateway Resources 169 m connector is the name of the connector OA ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH m vlan_ID is the VLAN identifier NO or 2 to 4094 Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN m p_key is the partition identifier 1 to 7fff or default For example to associate connector 1A ETH 3 with VLAN 3 and default partition key FabMan gateway_name gt createvlan 1A ETH 3 vlan 3 pkey default FabMan gateway_name gt 3 If you are associating VLANs with the Solaris VNICs perform these steps a Set the maximum number of VNICs per host FabMan gateway_name gt sethostvniclimit connector host hostname port port limit numb
81. Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 SS KG S u n Part No E36260 01 a March 2013 Revision A ORACLE Copyright 2013 Oracle and or its affiliates All rights reserved This software and related documentation are provided under a license agreement containing restrictions on use and disclosure and are protected by intellectual property laws Except as expressly permitted in your license agreement or allowed by law you may not use copy reproduce translate broadcast modify license transmit distribute exhibit perform publish or display any part in any form or by any means Reverse engineering disassembly or decompilation of this software unless required by law for interoperability is prohibited The information contained herein is subject to change without notice and is not warranted to be error free If you find any errors please report them to us in writing If this is software or related software documentation that is delivered to the U S Government or anyone licensing it on behalf of the U S Government the following notice is applicable U S GOVERNMENT END USERS Oracle programs including any operating system integrated software any pos installed on the hardware and or documentation delivered to U S Government end users are commercial computer software pursuant to the applicable Federal Acquisition Regulation and agency specific supplemental regulations As suc
82. System Name 208 Index 209 x Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Using This Documentation This administration guide provides detailed procedures that describe administration of the Sun Network QDR InfiniBand Gateway Switch Firmware Version 2 1 from Oracle This document is written for technicians system administrators and users who have advanced experience administrating InfiniBand fabric hardware a Product Notes on page xi m Related Documentation on page xi a Feedback on page xii m Access to Oracle Support on page xii Product Notes For late breaking information an known issues about this product refer to the product notes at http docs oracle com cd E36256_01 Related Documentation Documentation Sun Network QDR InfiniBand Gateway Switch Firmware Version 2 1 Links http docs oracle com cd E36256_01 xi Documentation Links Oracle Solaris 11 OS Oracle Integrated Lights Out Manager ILOM 3 0 All Oracle products http http BEED www oracle com goto Solarisi11 docs docs oracle com cd E19860 01 docs oracle com Feedback Provide feedback on this documentation at http www oracle com goto docfeedback Access to Oracle Support Oracle customers have access to electronic support through My Oracle Support For information visit http www oracle c
83. TE ACTIVE OVER net6 ip down no Note You must create the IP interface before an IP address is assigned to it 5 Assign an IP address to the IP interface ipadm create addr T static a I P_address 24 IP_interface v4 where m IP_address is the IP address to be assigned to the IP interface m IP_interface is the interface corresponding to the data link and VIOA For example to assign the IP address of 192 168 17 24 to IP interface net ipadm create addr T static a 192 168 17 24 24 net6 v4 168 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 6 Verify that the IP address was properly assigned ipadm show addr IP_interface v4 where IP_interface is the interface corresponding to the data link and VIOA For example ipadm show addr net6 v4 ADDROBJ TYPE STATE ADDR net6 v4 static ok 192 168 17 24 24 Related Information dladm command man page ipadm command man page m Ethernet Over InfiniBand Overview Oracle Solaris on page 161 m Verify That the Ethernet Over InfiniBand Driver Is Installed Oracle Solaris on page 162 m Install the Ethernet Over InfiniBand Driver Oracle Solaris on page 162 m Set Up VIOAs Oracle Solaris on page 163 m Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 V Create Oracle Solaris VNICs With VLAN Tagging Oracle
84. _Key for 14 switch chips without a corresponding Subnet Manager and set the replication password Related Information m Secret M_Key Overview on page 64 m Prepare for Secret M_Key Functionality on page 66 m Configuring Secret M_Keys on page 67 m Set the Replication Password on page 76 m Enable or Disable Secret M_Key Persistence on page 76 m Enable or Disable Secret M_Key Functionality on page 77 m Considering Situations In Regard to Secret M_Key on page 78 Vv Prepare for Secret M_Key Functionality 1 Update the smnodes list for all smnodes in the InfiniBand fabric See Update the smnodes List on page 134 2 Option Configure IPoIB functionality See Create the IPoIB Interface on page 135 3 Create and propagate the fabric element configuration See Create a Fabric Element Configuration on page 120 4 Set the replication password See Set the Replication Password on page 76 5 Enable M_Key persistence See Enable or Disable Secret M_Key Persistence on page 76 6 Enable secret M_Key functionality See Enable or Disable Secret M_Key Functionality on page 77 66 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 7 Configure the secret M_Keys See Configuring Secret M_Keys on page 67 Related Information Secret M_Key Overview on page 64 smsubnetprotection Command
85. _name gt smpartition create n testpartition pkey 5 m full FabMan gateway_name gt 6 Verify the partition configuration FabMan gateway_name gt smpartition list modified Sun DCS IB partition config file This file is generated do not edit yversion_number 13 Default 0x7ff ipoib ALL_CAS full ALL_SWITCHES full SELF full SUN_DCS 0x0001 ipoib ALL SWITCHES full testpartition 0x0005 defmember full FabMan gateway_name gt 7 Consider your next steps Administering the InfiniBand Fabric 111 a If you want to add GUIDs see Add or Remove a Port From a Partition on page 112 a If you want to modify the configuration see Modify a Partition or Port on page 114 a If you want to delete the configuration altogether see Delete a Partition on page 116 m Otherwise go to Step 8 8 End the configuration session and commit the configuration to the active partition configuration FabMan gateway_name gt smpartition commit FabMan gateway_name gt Related Information m Gateway Reference smpartition command a Display the InfiniBand Fabric Partition Configuration on page 93 m smpartition Command on page 106 a smpartition Command Guidelines on page 107 m Partitions and P_Keys on page 108 m Determine the Partitions and P_Keys on page 109 m Add or Remove a Port From a Partition on page 112 a Modify a Partition or P
86. _name gt smpartition remove n testpartition port 00212800013e9314 FabMan gateway_name gt 4 Verify the partition configuration FabMan gateway_name gt smpartition list modified Sun DCS IB partition config file This file is generated do not edit version_number 13 Default 0x7ff ipoib ALL _CAS full ALL SWITCHES full SELF full SUN_DCS 0x0001 ipoib ALL_SWITCHES full testpartition 0x0005 defmember full Administering the InfiniBand Fabric 113 0x00212800013e9313 0x00212800013e93 7 FabMan gateway_name gt 5 End the configuration session and commit the configuration to the active partition configuration FabMan gateway_name gt smpartition commit FabMan gateway_name gt Related Information m Gateway Reference smpartition command a smpartition Command on page 106 a smpartition Command Guidelines on page 107 a Partitions and P_Keys on page 108 m Determine the Partitions and P_Keys on page 109 m Create a User Partition on page 110 m Modify a Partition or Port on page 114 m Delete a Partition on page 116 m Remove User Partitions for Firmware Downgrade on page 117 V Modify a Partition or Port You can modify the configuration of the partition or the membership of the ports with the smpartition modify command 1 Initiate a partition configuration session on the management controller FabMan
87. a on page 23 Vv Display Free Memory and File System Space The showfree command gives output similar to the Linux free and df commands and can help with drive housekeeping chores On the management controller type FabMan gateway_name gt showfree md Memory 305 6 MB 61 3 oe Filesystems dev hda5 dev hda3 var log dev hda2 config tmpfs dev shm tmpfs tmp FabMan gateway_name gt Administering the Chassis 25 Related Information m Gateway Reference showfree command m Display Gateway Environmental and Operational Data on page 23 Vv Display Chassis FRU ID The showfruinfo command displays gateway chassis FRU ID information On the management controller type 26 FabMan gateway_name gt showfruinfo Sun_Man1R UNIX_Timestamp32 Sun_Fru_Description Vendor_ID_Code Vendor_ID_Code_Source Vendor_Name_And_Site_Location Sun_Part_Number Sun_Serial_Number Serial_Number_Format Initial_HW_Dash_Level Initial_HW_Rev_Level Sun_Fru_Shortname Sun_Hazard_Class_Code Sun_SpecPartNo Sun_FRU_LabelR Sun_Serial_Number FRU_Part_Dash_Number FabMan gateway_name gt Related Information Fri Dec 14 01 24 28 2012 ASSY NM2 GW 11 E1 01 4577 CELESTICA CORP 5111402 0110SJC 1010NG0040 4V3F1 2Y2W2X4S 03 50 NM2 gateway Y 885 1655 01 SAN JOSE CA US 1013AK208D 541 4188 01 m Gateway Reference showfruinfo command a Display Power Supply FRU ID on page
88. age 157 m Configure and Create VNICs for Host Manual Mode Linux on page 158 Vv Create the Central Configuration File Linux The etc infiniband m1x4_vnic conf file is shared by all hosts in the InfiniBand fabric You create the configuration file from the host MAC address file 1 Use the host MAC address list that you created previously See Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 2 Open a text editor to create the configuration file 154 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 3 Create an entry in the configuration file with this format name name mac mac ib_port device port vid vlan_ID vnic_id number bx string eport connector where m name is ethXX and XX is column one of the host MAC address file mac is column two of the host MAC address list m device is column four of the host MAC address list m port is column five of the host MAC address list a vlan_ID is column seven of the host MAC address list Note The vid parameter is optional m number is column one of the host MAC address list m string is column six of the host MAC address list m connector is column eight of the host MAC address list For example name ethl mac 00 30 48 7d de e4 ib_port m1x4_0 1 vid 0 vnic_id 1 bx 0003ba000100c70b eport 0A ETH 1 4 5 6 Repeat Step 3 for all rows of th
89. ame of the first connector to be added to the LAG 0A ETH 1 to OA ETH 4 and 1A ETH 1 to 1A ETH 4 Administering Gateway Resources 187 m connectorN is the name of the last connector to be added to the LAG 0A ETH 1 to OA ETH 4 and 1A ETH 1 to 1A ETH 4 For example FabMan gateway_name gt addlagport 0A12 OA ETH 3 lag updated FabMan gateway_name gt 2 Or to delete a connector from a LAG type FabMan gateway_name gt dellagport lagname connector1 connector2 connectorN where m lJagname is the identifier of the LAG m connector1 is the name of the first connector to be removed from the LAG OA ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 m connectorN is the name of the last connector to be removed from the LAG OA ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 For example FabMan gateway_name gt dellagport 0A12 OA ETH 3 lag updated FabMan gateway_name gt Related Information m Gateway Reference dellagport command a Enable LAG Mode on page 184 m Create LAGs on page 185 m Delete a LAG on page 188 a Disable LAG Mode on page 189 V Delete a LAG When a LAG is no longer needed or its connectors must be re purposed you can delete the LAG Note This same procedure deletes a Global Gateway 188 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 On the management contro
90. anagement Key on page 60 m Set the Subnet Manager Routing Algorithm on page 61 m Disable the Subnet Manager on page 63 Disable the Subnet Manager If your InfiniBand fabric has too many Subnet Managers you can disable and stop the Subnet Manager within the management controller by using the disablesm command On the management controller type FabMan gateway_name gt disablesm Stopping partitiond daemon OK Stopping IB Subnet Manager OK FabMan gateway_name gt Related Information m Gateway Reference disablesm command m Enable the Subnet Manager on page 62 m Set the Subnet Manager Priority on page 56 m Set the Subnet Manager Prefix on page 57 m Enable Subnet Manager Controlled Handover on page 58 m Set the Subnet Manager Readable Management Key on page 59 m Clear the Subnet Manager Readable Management Key on page 60 m Set the Subnet Manager Routing Algorithm on page 61 m Enable the Subnet Manager on page 62 Administering the Subnet Manager 63 Configuring Secure Fabric Management These topics enable you to increase security of your InfiniBand fabric m Secret M_Key Overview on page 64 a smsubnetprotection Command on page 65 m Prepare for Secret M_Key Functionality on page 66 a Configuring Secret M_Keys on page 67 m Set the Replication Password on page 76 a Enable or Disable Secret M_Key Pers
91. ar frequently in the documentation For example the VNIC specific configuration file contains the DEVICE parameter which format is described as DEVICE name where name is suggested to be ethXX and XX is a user specified sequence number VNIC names of eth0 eth1 to eth7 have been problematic with Linux kernels A solution is to use this format for the VNIC name ethgateway_instance_VNIC_instance where m gateway_instance is the gateway instance number as returned by the showgwconfig command m VNIC_instance is a sequential VNIC ID number assigned to the VNIC You can view the VNIC ID numbers with the showvnics command An example VNIC name might be eth03_03 Administering Gateway Resources 147 Using this new format the filename for the VNIC specific configuration file becomes ifcfg ethgateway_instance_VNIC_instance For the previous example the configuration filename is ifcfg eth03_03 For consistency the VNIC specific configuration file must contain these two parameters DEVICE name HWADDR mac where m name is ethgateway_instance_VNIC_instance as described previously m mac is the MAC assigned to the VNIC Related Information m Gateway Manual Mode Overview Linux on page 143 m Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 a Determine VLAN Associations for Gateway Manual Mode Linux on page 145 m Configure and Creat
92. art Fil FabMan gateway_name gt Related Information m Gateway Reference smpquery command m Display Low Level Detailed Information About a Port on page 91 Vv Display Low Level Detailed Information About a Port The smpquery command can provide very detailed information about a port On the management controller type FabMan gateway_name gt smpquery portinfo lid port where m lid is the LID of the node Administering the InfiniBand Fabric 91 port is the port of the node For example to see detailed information about port 23 on the switch with LID 15 type Note The output in the example is just a portion of the full output Related Information a Gateway Reference smpquery command FabMan gateway_name gt smpquery portinfo 15 23 Port info Lid 15 port 23 MEEV a ac a eok Queer oe ae adea aia eh ene ae ee i 0x0000000000000000 GLAOPREELH oeoa e sae eh oe eae ee as 0x0000000000000000 ao Kear ig eee a eee em eee eee 0x0000 SMELA 6 2 naa dee eae bead eee eee tad 0x0000 CapMask retires me aed eo eee ak 0x0 Diag COGS cadre ie tee e pai thane a 0x0000 MkeyLeasePeriod 2200 0 HO CAMP ORES sre e kpi aE ie Shee 0 LinkWidthEnabled 1X or 4X LinkWidthSupported 1X or 4X LinkWidthActive 4X LinkSpeedSupported 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LINnkState csucw det hee ed tte eee Active PhysLinkS
93. ateway 191 conditions for gateway manual mode 147 configuration parameters 144 152 controlling 190 creating 142 148 195 gateway manual mode 146 host manual mode 158 deleting 201 disabling 200 displaying 176 enabling 199 modifying 197 VNIC specific configuration file 151 Index 217 218 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013
94. ateway_name gt setsmmkey none Current SM settings smpriority 5 controlled_handover TRUE subnet_prefix Oxabbababe M_Key None Routing engine FatTree FabMan gateway_name gt 3 Enable the Subnet Manager See Enable the Subnet Manager on page 62 4 Repeat Step 1 to Step 3 for all Subnet Managers requiring the change Related Information m Gateway Reference setsmmkey command m Set the Subnet Manager Readable Management Key on page 59 m Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 m Set the Subnet Manager Priority on page 56 m Set the Subnet Manager Prefix on page 57 m Enable Subnet Manager Controlled Handover on page 58 m Set the Subnet Manager Routing Algorithm on page 61 m Enable the Subnet Manager on page 62 m Disable the Subnet Manager on page 63 Set the Subnet Manager Routing Algorithm The default routing algorithm is Fat Tree To change to the Min Hop routing algorithm use the setsmrouting command 1 On the management controller disable the Subnet Manager See Disable the Subnet Manager on page 63 Administering the Subnet Manager 61 2 Set the routing algorithm FabMan gateway_name gt setsmrouting fattree minhop For example FabMan gateway_name gt setsmrouting minhop Current SM settings smpriority 5 controlled_handover TRUE subnet_prefix
95. ation dans des conditions optimales de s curit Oracle Corporation et ses affili s d clinent toute responsabilit quant aux dommages caus s par l utilisation de ce logiciel ou mat riel pour ce type d applications Oracle et Java sont des marques d pos es d Oracle Corporation et ou de ses affili s Tout autre nom mentionn peut correspondre a des marques appartenant d autres propri taires qu Oracle Intel et Intel Xeon sont des marques ou des marques d pos es d Intel Corporation Toutes les marques SPARC sont utilis es sous licence et sont des marques ou des marques d pos es de SPARC International Inc AMD Opteron le logo AMD et le logo AMD Opteron sont des marques ou des marques d pos es d Advanced Micro Devices UNIX est une marque d pos e d The Open Group Ce logiciel ou mat riel et la documentation qui l accompagne peuvent fournir des informations ou des liens donnant acc s a des contenus des produits et des services manant de tiers Oracle Corporation et ses affili s d clinent toute responsabilit ou garantie expresse quant aux contenus produits ou services manant de tiers En aucun cas Oracle Corporation et ses affili s ne sauraient tre tenus pour responsables des pertes subies des co ts occasionn s ou des dommages caus s par l acc s a des contenus produits ou services tiers ou a leur utilisation Qa m Adobe PostScript Contents Using This Documentation xi Troubleshooting the Gatew
96. ation on the Gateway on page 191 m Create VLANs on page 192 m Delete VLANs on page 194 m Create VNICs on page 195 m Modify VNICs on page 197 a Enable VNICs on page 199 a Disable VNICs on page 200 m Delete VNICs on page 201 Related Information m Controlling LAGs on page 183 Controlling Gateway Ports and Parameters on page 202 m Controlling the Chassis on page 28 m Controlling the I4 Switch Chip on page 40 m Controlling the InfiniBand Fabric on page 97 m Controlling the Subnet Manager on page 56 a Installing Gateway Supportive Software Linux on page 137 m Creating VNICs Under Gateway Manual Mode Linux on page 142 m Creating VNICs Under Host Manual Mode Linux on page 148 m Creating Virtual IO Adapters Oracle Solaris on page 160 m Monitoring Gateway Resources on page 172 190 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 VNIC Allocation on the Gateway By default VNICs are created and allocated by the management controller of the gateway using Gateway Manual Mode The createvnic command creates the VNICs and the resulting VNIC resources are controlled by the gateway Alternatively individual hosts can create and allocate VNICs on the gateway There are two ways to enable this functionality m sethostvniclimit Running the sethostvniclimit command on t
97. ay 1 Gateway Hardware Problems 1 InfiniBand Fabric Problems 3 Network Management Troubleshooting Guidelines 6 Understanding Routing Through the Gateway 7 Switch Chip Port to QSFP Connectors and Link LED Routes 8 QSFP Connectors and Link LEDs to Switch Chip Port Routes 8 Signal Route Through the Gateway 9 Switch GUIDs Overview 10 Understanding Administrative Commands 13 Configuration Overview 13 Oracle ILOM Command Overview 16 Hardware Command Overview 16 InfiniBand Command Overview 17 Administering the Chassis 19 Monitoring the Chassis 19 v Display Gateway General Health 20 v Display Power Supply Status 20 v Check Board Level Voltages 21 Display Internal Temperatures 22 Display Fan Status 22 Display Gateway Environmental and Operational Data 23 Display Flash Drive Information 24 Display Free Memory and File System Space 25 Display Chassis FRU ID 26 Display Power Supply FRU ID 26 DENE SEEE SEEE GE EE E Display Gateway Firmware Versions 27 Oracle Auto Service Request 28 Controlling the Chassis 28 v Restart the Management Controller 29 v Change the root User Password 29 Administering the 14 Switch Chip 31 Monitoring the I4 Switch Chip 31 v Display the Switch Chip Port to QSFP Connector Mapping 32 v Locate a InfiniBand Device and Connector From the GUID and Port 33 Locate a Switch Chip or Connector From the GUID and Port 33 Display Switch Chip Boot Status 34 Display Link Status 35 Display Connectors Configure
98. be the routing through the gateway The first table maps the switch chip port to a QSFP connector The second table provides a reverse map When command output provides a switch chip port you can use these tables to determine the route that link is following m Switch Chip Port to QSFP Connectors and Link LED Routes on page 8 m QSFP Connectors and Link LEDs to Switch Chip Port Routes on page 8 m Signal Route Through the Gateway on page 9 Related Information m Gateway Hardware Problems on page 1 m InfiniBand Fabric Problems on page 3 m Network Management Troubleshooting Guidelines on page 6 m Switch GUIDs Overview on page 10 Troubleshooting the Gateway 7 Switch Chip Port to QSFP Connectors and Link LED Routes Port Connector Port Connector Port Connector Port Connector 1 1A ETH 1 P1 10 13B 19 OB 28 4A 1A ETH 2 P2 2 1A ETH 3 P3 11 12A 20 0A 29 5B 1A ETH 4 P4 3 0A ETH 1 P1 12 12B 21 1B 30 5A OA ETH 2 P2 4 OA ETH 3 P3 13 9B 22 1A 31 8A 0A ETH 4 P4 5 15A 14 9A 23 2B 32 8B 6 15B 15 10B 24 2A 33 7A 7 14A 16 10A 25 3B 34 7B 8 14B 17 11B 26 3A 35 6A 9 13A 18 11A 27 4B 36 6B 8 Related Information m Display the Switch Chip Port to QSFP Connector Mapping on page 32 a Display Link Status on page 35 m Display Switch Chip Port Status on page 37 m QSFP Connectors and Link LEDs to Switch Chip Port Routes on page 8 m Signal Route Through the Gate
99. ble to access the management controller restart the management controller See Restart the Management Controller on page 29 If you are unable to access the management controller power cycle the gateway Refer to Gateway Service removing the gateway from the rack Refer to Gateway Installation installing the gateway into the rack Reperform the firmware upgrade Refer to Gateway Remote Management upgrading the gateway firmware Troubleshooting the Gateway Situation Corrective Steps Performance of the InfiniBand fabric seems diminished An InfiniBand Link LED is blinking 1 Determine if there are errors or problems with the InfiniBand fabric See Perform Comprehensive Diagnostics for the Entire Fabric on page 98 Find 1x SDR or DDR Links in the Fabric on page 100 Determine Which Links Are Experiencing Significant Errors on page 100 2 Locate the affected nodes by the GUID provided in the output of the ibdiagnet command See Locate a Switch Chip or Connector From the GUID and Port on page 33 3 If the problem is at a cable connection swap the suspect cable with a known good cable or reconnect the cable to a known good remote port and repeat Step 1 Refer to Gateway Service servicing data cables 4 If the problem still remains at the cable connection disable and re enable the respective port and repeat Step 1 See Disable or Enable an External Port on page
100. bnet Manager on page 62 Related Information m Gateway Reference setcontrolledhandover command m Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 m Set the Subnet Manager Priority on page 56 m Set the Subnet Manager Prefix on page 57 m Set the Subnet Manager Readable Management Key on page 59 m Clear the Subnet Manager Readable Management Key on page 60 m Set the Subnet Manager Routing Algorithm on page 61 m Enable the Subnet Manager on page 62 m Disable the Subnet Manager on page 63 Set the Subnet Manager Readable Management Key For added security an easily readable management key or readable M_Key is used by the Subnet Manager Agent to initialize and configure network nodes Refer to Gateway Reference setsmmkey command for more information Note All Subnet Managers in the InfiniBand fabric must be configured with the same readable M_Key value Note For a more secure fabric configure secret M_Key functionality See Configuring Secure Fabric Management on page 64 Secret M_Key functionality overrides readable M_Key functionality 1 On the management controller disable the Subnet Manager See Disable the Subnet Manager on page 63 Administering the Subnet Manager 59 2 Set the readable management key FabMan gateway_name gt setsmmkey m_key where m_k
101. ch to a persistent value not in the secret M_Key configuration you isolate the gateway or switch from the fabric 1 Log in to the management controller of the gateway or switch to receive the local secret M_Key 2 Set the local secret M_Key FabMan gateway_name gt smsubnetprotection setlocalsecretmkey m_key where m_key is the management key 16 hexadecimal digits For example FabMan gateway_name gt smsubnetprotection setlocalsecretmkey 0x00abcdefabcdef04 Local mkey Oxabcdefabcdef04 Attribute secret FabMan gateway_name gt 3 Optional If you want to isolate the gateway or switch set the secret M_Key to persist and restart the management controller See m Enable or Disable Secret M_Key Persistence on page 76 m Restart the Management Controller on page 29 Related Information m Gateway Reference smsubnetprotection command m Create the Secret M_Key Configuration on page 67 m Add Secret M_Keys to the Configuration on page 69 Administering the Subnet Manager 75 Set the Secret M_Key to Use on page 72 Delete Secret M_Keys From the Configuration on page 73 V Set the Replication Password For further security a replication password is used to authenticate secret M_Key configurations distributed by the master Subnet Manager On the management controller that hosts the master Subnet Manager type FabMan gateway_name gt smsubnetprotection setreplica
102. cname 123 45 67 89 sw36gw gateway test 123 45 67 90 sw36 spineSwitch no test FabMan gateway_name gt 4 End the configuration session and commit to the new configuration FabMan gateway_name gt dconfig complete fabric config FabMan gateway_name gt 5 Perform Step 1 through Step 4 on the management controllers of all elements of the fabric Related Information Gateway Reference fAconfig command Display the InfiniBand Fabric Configuration on page 94 Fabric Director and Fabric Elements on page 125 fdconfig Command Overview on page 126 Create a Fabric Configuration on page 127 Add or Remove an Element From the Fabric Configuration on page 129 Performing Supportive Tasks These tasks support the extended functionality Create the smnodes List on page 133 Update the smnodes List on page 134 Create the IPoIB Interface on page 135 Delete the IPoIB Interface on page 136 Related Information Investigating Nodes on page 98 Controlling Nodes on page 101 Partitioning the InfiniBand Fabric on page 105 Correlating Fabric Elements on page 119 132 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 m Configuring the Fabric Director Node List on page 125 V Create the smnodes List If you are partitioning your InfiniBand fabric a list of valid Subnet Manager nodes must
103. command Note The IP addresses of the fabric element configuration file must be in synchronization with IP addresses of the smnodes list If you update the smnodes list you must update the fabric element configuration file Related Information m Gateway Reference createfabric command a Display the Fabric Element Configuration on page 96 m Create a Fabric Element Configuration on page 120 a Add or Delete an Element From the Configuration on page 122 a Modify an Element of the Configuration on page 123 Vv Create a Fabric Element Configuration 1 Verify the smnodes list See Display the smnodes List on page 95 2 Initiate a fabric element configuration session on the management controller FabMan gateway_name gt createfabric start empty FabMan gateway_name gt Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 3 Add a fabric element to the configuration FabMan gateway_name gt createfabric add element name hostname ilomIp IP_address redundantIP P_address where m hostname is the host name of the element m IP_address is the IP address for the element recognized by Oracle ILOM or IPoIB For example FabMan gateway_name gt createfabric add element name mnm12 gw 1 ilomIp 123 45 67 89 Element added FabMan gateway_name gt 4 Repeat Step 3 for any additional fabric elements you want to add to
104. command displays the node to node connectivity The output of the command depends on the size of your fabric 86 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Note You can use the ibnetdiscover command to determine the LIDs of the HCAs On the management controller type FabMan gateway_name gt ibnetdiscover Topology file generated on Sat Apr 13 22 28 55 2002 Max of 1 hops discovered Initiated from node 0021283a8389a0a0 port 0021283a8389a0a0 vendid 0x2c9 devid 0xbd36 sysimgguid 0x21283a8389a0a3 switchguid 0x21283a8389a0a0 21283a8389a0a0 Switch 36 S 0021283a8389a0a0 Sun DCS 36 QDR switch localhost enhanced port 0 lid 15 Ilmc 0 23 H 0003ba000100e388 2 3ba000100e38a mnm33 43 HCA 1 FabMan gateway_name gt lid 14 4xQDR Related Information m Gateway Reference ipnetdiscover command m Perform Comprehensive Diagnostics for the Entire Fabric on page 98 Vv Display a Route Through the Fabric Often you need to know the route between two nodes The ibtracert command can provide that information by displaying the GUIDs ports and LIDs of the nodes along the route On the management controller type FabMan gateway_name gt ibtracert slid dlid where m slid is the LID of the source node Administering the InfiniBand Fabric 87 m dlid is the LID of the destination node For examp
105. commands 17 Oracle ILOM 214 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 commands 16 secret M_Key 64 P P_Keys 108 109 partitioning 108 109 adding ports 112 creating configuration 110 deleting configuration 116 displaying configuration 93 firmware downgrades 117 InfiniBand fabric 105 modifying configuration 114 removing ports 112 starting configuration session 110 password 29 performing supportive tasks 132 perfquery command 89 poor performing links 100 ports autodisabling 43 counters 38 disabling 104 information 91 resetting 102 speed 103 status 37 power supply checking status 20 FRU ID 26 preparing secret M_Key 66 problem hardware 1 InfiniBand fabric 3 Q QSFP connector mapping from GUID 33 to switch chip port 32 R reboot command 29 recovery errors 100 removing element from configuration 129 ports from partition configuration 112 resetting port 102 restarting management controller 29 route QSFP to switch chip port 8 32 switch chip port to QSFP 8 32 through gateway 7 9 through the fabric 87 S secret M_Key 65 adding secret M_Keys 69 conditions for situations 78 configuring 67 configuring local 75 creating configuration 67 deleting secret M_Keys 73 disabling functionality 77 persistence 76 displaying 97 configuration 52 persistence 53 enabling functionality 77 persistence 76 f
106. d 90 91 smsubnetprotection add command 69 delete command 73 disablesecretmkey command 77 enablesecretmkey command 77 list active command 52 97 set current command 72 setlocalsecretmkey command 75 setreplicationpassword command 76 start command 67 smsubnetprotection command 65 status chassis 23 fan 22 flash drive 24 power supply 20 Subnet Manager 50 Subnet Manager added to secret M_Key fabric 78 clearing management key 60 configuration integrity 54 controlled handover state 51 controlling 56 disabling 63 displaying configuration integrity 54 location history 51 log 55 management key 51 prefix 51 priority 51 routing algorithm 51 status 50 enabling 62 management key 51 59 60 monitoring 49 setting controlled handover 58 management key 59 prefix 57 priority 56 routing algorithm 61 supportive tasks 132 switch 216 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 GUIDs 10 identity 84 switch chip checking boot status 34 port disabling 40 enabling 41 symbol errors 100 system name setting 208 T topology change 99 displaying detailed 86 simple 86 troubleshooting gateway 1 network management 6 V verifying EoIB driver installed 162 version command 27 VIOAs 161 creating 163 VLANs associations 145 controlling 190 creating 192 deleting 194 displaying 174 VNICs allocation on g
107. d With Autodisable Functionality 36 DENS SES SEE SEEE Display Switch Chip Port Status 37 v Display Switch Chip Port Counters 38 Controlling the 14 Switch Chip 40 v Disable a Switch Chip Port 40 v Enable a Switch Chip Port 41 Autodisabling Ports and Connectors 43 Autodisable Functionality 43 iv Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 v Add Connectors to the Autodisable List 44 v Reenable Autodisabled Connectors 45 v Delete Connectors From the Autodisable List 46 Administering the Subnet Manager 49 Monitoring the Subnet Manager 49 v v lt a lt v Display Subnet Manager Status 50 Display the Subnet Manager s Location History 51 Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm 51 Display the Secret M_Key Configuration 52 Verify M_Key Persistence 53 Verify the Subnet Manager Configuration Integrity 54 Display the Subnet Manager Log 55 Controlling the Subnet Manager 56 v 4 lt a v Set the Subnet Manager Priority 56 Set the Subnet Manager Prefix 57 Enable Subnet Manager Controlled Handover 58 Set the Subnet Manager Readable Management Key 59 Clear the Subnet Manager Readable Management Key 60 Set the Subnet Manager Routing Algorithm 61 Enable the Subnet Manager 62 Disable the Subnet Manager 63 Configuring Secure Fabric Management 64 Secret M_Key Overview 64 sms
108. d for external ports Use the disableswitchport and enableswitchport commands for internal ports when secret M_Key functionality is enabled See Disable a Switch Chip Port on page 40 and Enable a Switch Chip Port on page 41 On the management controller type FabMan gateway_name gt ibportstate M m_key lid port state where m m_key is the management key 16 hexadecimal digits optional 104 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m lid is the LID of the node m port is the port of the node m state is either disable or enable For example FabMan gateway_name gt ibportstate 15 23 disable Initial PortInfo Port info Lid 15 port 23 LiInkState isesi eee eek ee ee edn Active PhysLinkState 202200 LinkUp LinkWidthSupported 1X or 4X LinkWidthEnabled 1X or 4X LinkWidthActive 4X LinkSpeedSupported 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedEnabled 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedActive 00 ee eee 10 0 Gbps After PortInfo set Port info Lid 15 port 23 PhysSLinkState csobe sada eee ae eas Disabled FabMan gateway_name gt Related Information Gateway Reference ipportstate command Disable a Switch Chip Port on page 40 Disable or Enable an External Port on page 104 Reset a Port on page 1
109. d into each link The P al1 1 option returns all Performance Monitor counters that increment during the test respective to the GUID and port of the InfiniBand device 100 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 2 In the output of the ibdiagnet command search for the symbol_error_counter string That line contains the symbol error count in hexadecimal The preceding lines identify the node and port with the errors Symbol errors are minor errors If there are relatively few minor errors during the diagnostic they can be monitored Note According to InfiniBand specification 10E 12 BER the maximum allowable symbol error rate is 120 errors per hour 3 Also in the output of the ibdiagnet command search for the link_error_recovery_counter string That line contains the recovery error count in hexadecimal The preceding lines identify the node and port with the errors Recovery errors are major errors The respective links must be investigated for the cause of the rapid symbol error propagation Related Information m Gateway Reference ibdiagnet command m Find 1x SDR or DDR Links in the Fabric on page 100 m Display Counters for a Node on page 89 Controlling Nodes These tasks enable you to manage ports in the InfiniBand fabric m Clear Data and Error Counters on page 102 m Reset a Port on page 102 m Set Port Speed on page 103
110. d348 Ox3fc7e6ad701la8a2a S FabMan gateway_name gt Note The s in the Attribute column means that M_Key is in standby state and is to become the current secret M_Key 7 End the session and commit the secret M_Key configuration to the master Subnet Manager FabMan gateway_name gt smsubnetprotection commit FabMan gateway_name gt The configuration is distributed to the trusted standby Subnet Managers and the current secret M_Key to the InfiniBand Fabric Related Information m Gateway Reference smsubnetprotection command m Add Secret M_Keys to the Configuration on page 69 m Set the Secret M_Key to Use on page 72 m Delete Secret M_Keys From the Configuration on page 73 m Configure a Local Secret M_Key on page 75 V Add Secret M_Keys to the Configuration Note There is a maximum of 10 secret M_Keys for the configuration Administering the Subnet Manager 69 1 Initiate a secret M_Key configuration session on the management controller that hosts the master Subnet Manager FabMan gateway_name gt smsubnetprotection start addonly FabMan gateway_name gt 2 Display the active secret M_Key configuration See Display the Secret M_Key Configuration on page 52 3 Add an M_Key to the configuration FabMan gateway_name gt smsubnetprotection add m_key where m_key is the management key 16 hexadecimal digits For example FabMan gateway_name gt sms
111. de for Firmware Version 2 1 e March 2013 m Controlling the Subnet Manager on page 56 m Monitoring the Chassis on page 19 V Restart the Management Controller Should the management controller enter an indeterminate state you can restart it The Subnet Manager and other services are also restarted Additionally the links to the I4 switch chip are disabled and then re enabled when the management controller reboots Note This procedure severs any management console link to the management controller You must reaccess the management controller to regain administrative control At the Oracle ILOM prompt type gt reset SP Are you sure you want to reset SP y n y Performing reset on SP Broadcast message from root Wed Sep 5 03 15 55 2012 The system is going down for reboot NOW gt Connection to gateway_name closed by remote host Connection to gateway_name closed The management controller power cycles and reinitializes Related Information m Change the root User Password on page 29 V Change the root User Password The default password for the root user is changeme You can use the passwd command to make the password unique and more secure Note For instructions to change the Oracle ILOM user passwords refer to Gateway Remote Management changing Oracle ILOM user passwords 1 On the management controller log in as the root user Administering the Chassis 29 2 Type
112. dth Spe ed Enabled changed from LID 1 GUID 0x00212856d102c0a0 new CapabilityMask 0x4250084a Sep 05 04 49 30 993000 B6658B90 0x02 gt Fabric has 1 switches topology is not fat tree Falling back to default routing Sep 05 04 49 30 993000 B6658B90 0x01 gt ucast_mgr_route ftree cannot build lid matrices END FabMan gateway_name gt The output first displays the parameters of the Subnet Manager configuration file and then a chronological listing of Subnet Manager activity 2 Tap the space bar to display the next screen of the log 3 Press the Q key to quit Related Information m Gateway Reference showsmlog command m Display Subnet Manager Status on page 50 m Display the Subnet Manager s Location History on page 51 m Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 m Verify the Subnet Manager Configuration Integrity on page 54 Administering the Subnet Manager 55 Controlling the Subnet Manager You can enable the Subnet Manager with the enablesm command When the Subnet Manager starts it reads the configuration file for configuration information You can disable the Subnet Manager with the disablesm command These topics describe how to control the Subnet Manager Note To use the commands described in these topics you must access them through the SYS Fabric_Mgmt Linux shell targets of the Oracle
113. e edee ca faces chee e LinkUp LinkWidthSupported 1X or 4X LinkWidthEnabled 1X or 4X LinkWidthActive 0 4X LinkSpeedSupported 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedEnabled 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedActive 24 10 0 Gbps After PortInfo set t Port info DR path slid 65535 dlid 65535 0 port 14 PhysLinkState 2008 Disabled FabMan gateway_name gt Related Information m Gateway Reference disableswitchport command m Disable or Enable an External Port on page 104 m Enable a Switch Chip Port on page 41 m Autodisabling Ports and Connectors on page 43 Vv Enable a Switch Chip Port You can enable a disabled switch chip port with the enableswitchport command Administering the 14 Switch Chip 41 On the management controller type FabMan gateway_name gt enableswitchport reason reason Switch port where m reason is the reason for disabling the port Blacklist or Partition m port is the number of the port 1 to 36 Note If the port was disabled with the reason option as seen with the listlinkup command it can only be enabled with the same reason option Alternatively type FabMan gateway_name gt enableswitchport reason reason connector where connector is the number of the QSFP connector 0A to 15B For example FabMan gat
114. e 186 m Add or Delete Connectors From a LAG on page 187 m Delete a LAG on page 188 m Disable LAG Mode on page 189 Administering Gateway Resources 183 Related Information m Display LAG Information on page 173 m Controlling VLANs and VNICs on page 190 Controlling Gateway Ports and Parameters on page 202 m Controlling the Chassis on page 28 m Controlling the I4 Switch Chip on page 40 m Controlling the InfiniBand Fabric on page 97 m Controlling the Subnet Manager on page 56 a Installing Gateway Supportive Software Linux on page 137 m Creating VNICs Under Gateway Manual Mode Linux on page 142 m Creating VNICs Under Host Manual Mode Linux on page 148 m Creating Virtual IO Adapters Oracle Solaris on page 160 m Monitoring Gateway Resources on page 172 Y Enable LAG Mode Note LAG mode must be enabled on all InfiniBand gateways in the fabric before LAGs are created 1 On the management controller type FabMan gateway_name gt enablelagmode VNICs for host nodes with EoIB driver version that do not support LAG will go down if LAG mode is enabled Do you still want to enable LAG y n y Stopping Bridge Manager OK Starting Bridge Manager OK FabMan gateway_name gt 2 Repeat Step 1 for all InfiniBand gateways in the fabric Related Information m Gateway Reference enablelagmode command m C
115. e Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 Vv Configure IP Addresses on the Data Link Oracle Solaris 1 Become superuser of the Oracle Solaris 11 host 2 Verify that the data link corresponding to the activated VIOA has been created For example dladm show phys LINK MEDIA STATE SPEED DUPLEX DEVICE vneto Ethernet up 0 unknown vnet0 ibp0 Infiniband up 32000 unknown ibpo ibp1 Infiniband down 2000 unknown ibp1 net5 Ethernet up 10000 full eoib0 net4 Ethernet up 10000 full eoib1 net6 Ethernet unknown 10000 full eoib2 The dladm show phys command displays all data links corresponding to all hardware devices Similarly the dladm show link command displays the class MTU and state for all data links For example dladm show link LINK CLASS MTU STATE OVER vneto phys 1500 up ibpo phys 65520 up ibp1 Administering Gateway Resources 167 net5 phys 1500 up net4 phys 1500 up an net6 phys 1500 unknown 3 Create an IP interface on the data link ipadm create ip link where link is the name of the data link For example ipadm create ip net6 GI Note When you create the IP interface on the data link the respective VNIC STAT as seen with the showvnics command goes from WAIT IOA to UP 4 Verify that the interface was created ipadm show if IFNAME CLASS STA
116. e Solaris on page 162 3 Optional Install the driver Install the Ethernet Over InfiniBand Driver Oracle Solaris on page 162 4 Create the VIOAs Set Up VIOAs Oracle Solaris on page 163 5 Create IP interfaces and Configure IP Addresses on the Data Link Oracle assign IP addresses to the Solaris on page 167 VIOAs 6 Create Oracle Solaris VNICs Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 Related Information Installing Gateway Supportive Software Linux on page 137 Creating VNICs Under Gateway Manual Mode Linux on page 142 Creating VNICs Under Host Manual Mode Linux on page 148 Monitoring Gateway Resources on page 172 Controlling LAGs on page 183 Controlling VLANs and VNICs on page 190 Controlling Gateway Ports and Parameters on page 202 160 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Ethernet Over InfiniBand Overview Oracle Solaris The Ethernet over InfiniBand driver is available in current releases of the Oracle Solaris 11 image This driver supports the Data Link Provider Interface over all InfiniBand ports of an Oracle Solaris 11 host connected to the gateway The driver uses the IBA unreliable datagram mode to enable initialization gateway handshake heartbeat management frame transmit and receive functions multicast support and statistical reporting
117. e VNICs for Gateway Manual Mode Linux on page 146 Creating VNICs Under Host Manual Mode Linux These topics describe how to create VNICs under host manual mode m Host Manual Mode Overview Linux on page 149 m Central Configuration File Linux on page 150 a VNIC Specific Configuration File Linux on page 151 m Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 m Create the Central Configuration File Linux on page 154 m Create the VNIC Specific Configuration Files Linux on page 156 m mlx4_vnic_confd Daemon Linux on page 157 m Configure and Create VNICs for Host Manual Mode Linux on page 158 148 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information m Installing Gateway Supportive Software Linux on page 137 m Creating VNICs Under Gateway Manual Mode Linux on page 142 m Creating Virtual IO Adapters Oracle Solaris on page 160 m Monitoring Gateway Resources on page 172 m Controlling LAGs on page 183 m Controlling VLANs and VNICs on page 190 Controlling Gateway Ports and Parameters on page 202 Host Manual Mode Overview Linux Before creating VNICs in host manual mode you must install the BXOFED software onto the hosts Afterwards you create VNICs using static configuration files located on the hosts These confi
118. e host MAC address list Save the file with the name m1x4_vnic conf Copy the m1x4_vnic conf file to the etc infiniband directory of all hosts in the InfiniBand fabric Related Information Create the VNIC Specific Configuration Files Linux on page 156 Host Manual Mode Overview Linux on page 149 Central Configuration File Linux on page 150 VNIC Specific Configuration File Linux on page 151 Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 mlx4_vnic_confd Daemon Linux on page 157 Configure and Create VNICs for Host Manual Mode Linux on page 158 Administering Gateway Resources 155 Vv Create the VNIC Specific Configuration Files Linux Each host will have a VNIC specific configuration file for the VNICs it will host Each line of the MAC address file becomes a VNIC specific configuration file Repeat this procedure for each host in the InfiniBand fabric 1 Use the host MAC address list that you created previously See Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 2 Look in column three of the host MAC address list to find entries for the respective host Note the row numbers in column one 3 Open a text editor to create a configuration file 4 Enter text from the entry appropriate for the respective host into the file with this format DEVICE name HWADDR mac BOOTPROTO dhcp ONBOOT yes BX
119. entify which VLANs are associated with the gateway 10GbE connectors 1 On the management controller type FabMan gateway_name gt showvlan Connector LAG VLN EY EEF EEEF 174 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 OA ETH 2 3 ffff OA ETH 2 0 ffff OA ETH 3 3 FECE 0A ETH 3 0 FLEE 0A ETH 4 3 FEEF 0A ETH 4 0 ELEF 1A ETH 1 3 FEFE 1A ETH 1 0 ffff 1A ETH 2 3 EEEE 1A ETH 2 0 ffff 1A ETH 3 3 FELE 1A ETH 3 0 FEFEFE 1A ETH 4 3 FETE 1A ETH 4 0 ffff FabMan gateway_name gt Note A VLAN identifier VLN of 0 1 or NO indicates that there is no VLAN assigned Note Your output will differ from that in the example The showvlan command lists the connectors on the left and the associated VLANs and respective partition keys on the right 2 If the information displayed is incorrect modify the VLAN configuration See Controlling VLANs and VNICs on page 190 Related Information m Gateway Reference showvlan command a Controlling VLANs and VNICs on page 190 m Display LAG Information on page 173 m Display the VNICs on page 176 a Display IO Adapters on page 177 a Display Gateway Ethernet Port Information on page 178 m Display Gateway Port Information on page 179 m Display the Gateway Instance Number on page 181 m Display Information About the Gateway on page 182 Administering
120. er vlanlist vlan_ID1 vlan_IDn where connector is the name of the connector OA ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 hostname is the host name of the management controller or system name recognized by the BridgeX manager port is the port of the host a number is the maximum number of VNICs for the IOA a vlan_ID is the VLAN identifier NO or 2 to 4094 Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN For example to limit the host cupcake to 4 VNICs FabMan gateway_name gt sethostvniclimit OA ETH 1 host cupcake port 1 limit 4 vlanlist 3 FabMan gateway_name gt 170 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 b Repeat Step a for all hosts in the fabric that need to be regulated 4 On the Oracle Solaris 11 host create a Oracle Solaris VNIC on the data link dladm create vnic 1 IP interface m mac_address vw vlan_ID name where m IP_interface is the interface corresponding to the data link and VIOA m mac_address is the MAC address to be assigned to the VIOA m vlan_ID is the VLAN identifier NO or 2 to 4094 Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 i
121. ess append the name of the gateway connector where the host will physically connect to the 10GbE network to the right of the GUID or VLAN identifier if present Note Given a connector and vlan_id combination there is only one p_key permitted Administering Gateway Resources 153 9 Use the list for when you create VLANs and VNICs An example entry in the host MAC address list might look like this 1 00 30 48 7d de e4 cupcake mlx4_0 1 0003ba000100c70b 3 OA ETH 1 For this example m 1 is the entry number This will become the VNIC number and eth number m 00 30 48 7d de e4 is the MAC address a cupcake is the host name and is used to identify the location of the VNIC m mlx4_0 is the value of the hca_id field and is called the device name m 1 is the port of the HCA m 0003ba000100c70b is the HCA port GUID m 3 is the VLAN identifier m OA ETH 1 is the connector Related Information m Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 a Determine VLAN Associations for Gateway Manual Mode Linux on page 145 a Host Manual Mode Overview Linux on page 149 m Central Configuration File Linux on page 150 a VNIC Specific Configuration File Linux on page 151 m Create the Central Configuration File Linux on page 154 m Create the VNIC Specific Configuration Files Linux on page 156 m mlx4_vnic_confd Daemon Linux on p
122. etevnic 201 dellagport 187 disablegwport 204 disablelagmode 189 disablesm 63 disableswitchport 40 disablevnic 200 dladm 163 167 169 enablegwport 203 enablelagmode 184 enablesm 62 enableswitchport 41 45 enablevnic 199 env_test 23 fdconfig 126 abort 127 complete fabric config 127 define element 129 define fabric name 127 list current fabric config 94 redefine element 131 remove element 129 start fabric config empty 127 generatetopology 99 getfanspeed 22 getmaster 51 getportcounters 38 getportstatus 37 ibdiagnet 98 100 102 ibhosts 85 ibnetdiscover 86 ibportstate 88 102 103 104 ibswitches 84 ibtracert 87 ipadm 167 listlinkup 35 localmkeypersistence 53 76 matchtopology 99 modifyvnic 197 perfquery 89 reboot 29 setgwethport 178 205 setgwinstance 181 207 setgwsl 206 setgwsystemname 208 setsmmkey 59 60 setsmpriority 51 56 58 setsmrouting 61 setsubnetprefix 57 showdisk 24 showfree 25 showfruinfo 26 showgwconfig 182 showgwports 179 showioadapters 177 showlag 173 showpsufru 26 showsmlog 55 showtemps 22 showtopology 86 showunhealthy 20 showvlan 174 showvnics 176 smconfigtest 54 sminfo 50 smnodes add 133 delete 134 list 95 smpartition 106 add 112 create 110 delete 116 list 93 modify 114 remove 112 start 110 smpquery 90 91 smsubnetprotection 65 67 add 69 delete 73 disablesecretmkey 77 enablesecretmkey
123. eway MAC address file might look like this OA ETH 1 0003ba000100c70b 00 30 48 7d de e4 For this example m OA ETH 1 is the gateway connector m 0003ba000100c70b is the HCA port GUID m 00 30 48 7d de e4 is the MAC address 144 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information m Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 m Gateway Manual Mode Overview Linux on page 143 m Determine VLAN Associations for Gateway Manual Mode Linux on page 145 m Configure and Create VNICs for Gateway Manual Mode Linux on page 146 m VNIC Configuration on Linux Hosts in Gateway Manual Mode on page 147 Determine VLAN Associations for Gateway Manual Mode Linux You must associate a VNIC with a VLAN even if that VLAN identifier is NO You also must partition the InfiniBand fabric and provide a partition key for each VLAN If the VLAN identifier NO is to be associated with the P_Key default you need not create the default partition for it already exists Note The association of VLAN identifier NO to partition key default survives firmware upgrades 1 If you have not done already partition the InfiniBand fabric See Partitioning the InfiniBand Fabric on page 105 2 Use the smpartition command on the management controller to list the GUIDs associated with the partition and partition ke
124. eway_name gt enableswitchport Switch 14 Enable Switch port 14 Initial PortInfo Port info DR path slid 65535 dlid 65535 0 port 14 PinkS tates sess see Sea de BS ad wk Down PhysLinkState Disabled LinkWidthSupported 1X or 4X LinkWidthEnabled 1X or 4X LinkWidthActive 4X LinkSpeedSupported 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedEnabled 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedActive 0 0200 2 5 Gbps After PortInfo set Port info DR path slid 65535 dlid 65535 0 port 14 DIRS tACS Sasana a eee ae OR eee Down PHYSLInkState ty sae ee ee ee eS PortConfigurationTraining FabMan gateway_name gt Related Information Gateway Reference enableswitchport command Disable or Enable an External Port on page 104 Disable a Switch Chip Port on page 40 Reenable Autodisabled Connectors on page 45 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Autodisabling Ports and Connectors These topics instruct how to manage autodisabled ports and connectors m Autodisable Functionality on page 43 m Add Connectors to the Autodisable List on page 44 m Reenable Autodisabled Connectors on page 45 m Delete Connectors From the Autodisable List on page 46 Related Information m Disable a Switch Chip Port on page
125. ey is the management key 16 hexadecimal digits For example FabMan gateway_name gt setsmmkey Oxabbababe0001 Current SM settings smpriority 5 controlled_handover TRUE subnet_prefix Oxabbababe M_Key Oxabbababe0001 Routing engine FatTree FabMan gateway_name gt 3 4 Enable the Subnet Manager See Enable the Subnet Manager on page 62 Repeat Step 1 to Step 3 for all Subnet Managers of the InfiniBand fabric Related Information Gateway Reference setsmmkey command Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 Set the Subnet Manager Priority on page 56 Set the Subnet Manager Prefix on page 57 Enable Subnet Manager Controlled Handover on page 58 Clear the Subnet Manager Readable Management Key on page 60 Set the Subnet Manager Routing Algorithm on page 61 Enable the Subnet Manager on page 62 Disable the Subnet Manager on page 63 Vv Clear the Subnet Manager Readable Management Key It might be necessary to remove the restrictions of the Subnet Manager readable management key This procedure describes how to do so 60 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 1 On the management controller disable the Subnet Manager See Disable the Subnet Manager on page 63 2 Clear the readable management key FabMan g
126. f 1A ETH 4 13 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 01 3 ffff 1A ETH 3 14 UP N 00 03 BA 00 01 00 E3 72 mnm34 60 0000 02 02 02 02 02 02 3 ffff 1A ETH 2 FabMan gateway_name gt 5 If you deleted the VNICs in order to delete VLANs delete the VLANs See Delete VLANs on page 194 Related Information m Gateway Reference deletevnic command m Gateway Reference showvnics command a VNIC Allocation on the Gateway on page 191 m Create VLANs on page 192 a Delete VLANs on page 194 m Create VNICs on page 195 a Enable VNICs on page 199 a Disable VNICs on page 200 m Display the VNICs on page 176 Controlling Gateway Ports and Parameters These topics describe how you can control the ports and parameters of the Sun Network QDR InfiniBand Gateway Switch from Oracle 202 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Note To use the commands described in these topics you must access them through the SYS Gateway_Mgmt or SYS Fabric_Mgmt Linux shell targets of the Oracle ILOM CLI m Enable a Gateway Ethernet Port on page 203 m Disable a Gateway Ethernet Port on page 204 m Set the Gateway Ethernet Port Parameters on page 205 m Set the Gateway Service Level on page 206 m Set the Gateway Instance Number on page 207 m Set the Gateway System Name on page 208 Related
127. f the node m port is the port of the node Note If a port value of 255 is specified for a switch node the counters are the total for all switch ports For example Note The output in the example is just a portion of the full output FabMan gateway_name gt perfquery 15 23 Port counters Lid 15 port 23 POLCSEL SCE i ressadas aa a See eels wd 23 CounterSelect 22 Ox1b01 SYMDOLEFTOLS ios Ieis eee se de Bader dec 0 VLLS DEOp pei sds da apa he a Saas ae 0 MMCDACAs esde ika RE aR eee OS 20232 ROVDACAt ic ke ka oe Swed eos Be aS 20232 XMEPKES ee diene be away E eee ee 281 REVPKES 8 4 aaa E E ed Sa math ede 281 FabMan gateway_name gt Related Information C Gateway Reference perfquery command m Display Switch Chip Port Counters on page 38 m Clear Data and Error Counters on page 102 Vv Display Low Level Detailed Information About a Node If intensive troubleshooting is necessary to resolve a problem the smpquery command can provide very detailed information about a node 90 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 On the management controller type FabMan gateway_name gt smpquery switchinfo lid where lid is the LID of the node For example to see detailed information about a switch with LID 15 type FabMan gateway_name gt smpquery switchinfo 15 Switch info Lid 15 OutboundP
128. file This file is generated do not edit secretmkey enabled nodeid o04nm2 gw 6 time 15 Sep 04 03 00 checksum 0 commit_number 4 Mkey Untrusted Mkey Smkey Attribute oF OH OHHH 0x00abcdefabcdef01 Oxlaa45124fee612ae Ox15fc26aea300 831 0x00abcdefabcdef02 0x4ccd8230de6cd348 Ox3fc7e6ad701la8a2a 0x00abcdefabcdef03 0x9baaldebcc74de5e 0x1b253003600d137b FabMan gateway_name gt In the output the Attribute heading identifies m C The current secret M_Key m S The standby secret M_Key if any Related Information m Gateway Reference smsubnetprotection command m Configuring Secret M_Keys on page 67 V Verify M_Key Persistence You can check whether the M_Key survives a local power cycle restart or reboot 1 On the management controller type FabMan gateway_name gt localmkeypersistence show Local M_Key persistence is disabled No local M Key is set before a Subnet Manager is operational FabMan gateway_name gt Administering the Subnet Manager 53 2 If the information displayed is incorrect change it See Enable or Disable Secret M_Key Persistence on page 76 Related Information m Gateway Reference localmkeypersistence command a Enable or Disable Secret M_Key Persistence on page 76 V Verify the Subnet Manager Configuration Integrity The smconfigtest command performs a check of the Subnet Manager s configuration and verifies the integrity Note For the spine opt
129. fy ALL CAS All CAs in the InfiniBand fabric ALL_SWITCHES All switches a ALL ROUTERS All routers m member is the membership type full limited or both for the port For example to configure the default partition for limited membership for CAs and no IP over InfiniBand support FabMan gateway_name gt smpartition modify pkey 0x7fff port ALL_CAS flag m limited FabMan gateway_name gt Administering the InfiniBand Fabric 115 3 Verify the partition configuration FabMan gateway_name gt smpartition list modified Sun DCS IB partition config file This file is generated do not edit version_number 13 Default 0x7fff ipoib ALL _CAS full ALL_SWITCHES full SELF full SUN_DCS 0x0001 ipoib ALL _SWITCHES full testpartition 0x0005 defmember full ipoib 0x00212800013e9313 0x00212800013e93f7 4 End the configuration session and commit the configuration to the active partition configuration FabMan gateway_name gt FabMan gateway_name gt smpartition commit FabMan gateway_name gt Related Information m Gateway Reference smpartition command a smpartition Command on page 106 m smpartition Command Guidelines on page 107 m Partitions and P_Keys on page 108 a Determine the Partitions and P_Keys on page 109 a Create a User Partition on page 110 m Add or Remove a Port From a Partition o
130. gateway_name gt smpartition start FabMan gateway_name gt 2 Modify the partition or ports FabMan gateway_name gt smpartition modify n partition_name pkey p_key flag ipoib mtu mtu rate rate sl sl scope scope port port ALL_CAS ALL_SWITCHES ALL_ROUTERS m member where m partition_name is an alphanumeric tag to the InfiniBand partition optional m p_key is the partition key 1 to 7 ff 114 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m mtu is the number that maps to the actual MTU 1 to 5 mtu Number 1 2 3 4 5 MTU Value 256 512 1024 2048 4096 m rate is the number that maps to the actual throughput of a link link width link speed 2 to 10 rate Number 2 3 4 5 6 7 8 9 10 Rate Value in Gbps 2 5 10 30 5 20 40 60 80 120 m sl is the service level 0 to 15 Note Use service level 1 s1 1 only for low latency high priority small message low bandwidth traffic Use other service levels for regular high bandwidth traffic m scope is the multicast address scope value 1 to 14 Note The mtu rate s1 and scope parameters are for the multicast group created when ipoib IP over InfiniBand is configured for the partition Typically these values are not specified as the defaults are sufficient for the fabric configuration m port is the GUID of the port or the special parameter to modi
131. gement controller recognizes a system name that is independent of the management controller s host name If no system name is configured the host name is used The system name is a maximum of 19 characters 1 On the management controller type FabMan gateway_name gt setgwsystemname systemname where systemname is the BridgeX manager s system name For example FabMan gateway_name gt setgwsystemname brooklyn Stopping Bridge Manager OK Starting Bridge Manager OK FabMan gateway_name gt setgwsystemname list BXM system name set to myname FabMan gateway_name gt 2 Use the list option of the setgwsystemname command to verify the name configuration FabMan gateway_name gt setgwsystemname list BXM system name set to brooklyn FabMan gateway_name gt Related Information m Gateway Reference setgwsystemname command a Enable a Gateway Ethernet Port on page 203 a Disable a Gateway Ethernet Port on page 204 m Set the Gateway Ethernet Port Parameters on page 205 m Set the Gateway Service Level on page 206 m Set the Gateway Instance Number on page 207 a Display Information About the Gateway on page 182 208 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Index A acquiring BXOFED software 138 ConnectX 2 firmware 141 adding autodisabled connectors 44 connectors to LAGs 187 element t
132. grade Guidelines on page 80 Related Information a Secret M_Key Overview on page 64 a smsubnetprotection Command on page 65 m Prepare for Secret M_Key Functionality on page 66 m Configuring Secret M_Keys on page 67 m Set the Replication Password on page 76 m Enable or Disable Secret M_Key Persistence on page 76 a Enable or Disable Secret M_Key Functionality on page 77 V Add a New Subnet Manager to the Secret M_Key Fabric When you add a switch or gateway to your existing fabric you might also enable a Subnet Manager on that switch Alternatively you might enable a previously disabled Subnet Manager In this situation follow these instructions 1 If not installed install the switch or gateway hosting the Subnet Manager Refer to Switch Installation or Gateway Installation Note Do not cable signal or InfiniBand cables to the switch or gateway at this time 2 If the firmware is less than version 2 1 upgrade it Refer to Switch Remote Management or Gateway Remote Management upgrading the firmware 78 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Verify that the Subnet Manager is disabled See Disable the Subnet Manager on page 63 Set the Subnet Manager Priority to a low value so that it is a standby Subnet Manager See Set the Subnet Manager Priority on page 56 Enable the Subnet Manager
133. guration 127 v Add or Remove an Element From the Fabric Configuration 129 v Modify an Element of the Fabric Configuration 131 Performing Supportive Tasks 132 Create the smnodes List 133 lt Update the smnodes List 134 v Create the IPoIB Interface 135 v Delete the IPoIB Interface 136 Administering Gateway Resources 137 Installing Gateway Supportive Software Linux 137 v Acquire the BXOFED Software Linux 138 v Install the BXOFED Software Linux 139 v Acquire the ConnectX 2 Firmware 141 Creating VNICs Under Gateway Manual Mode Linux 142 Gateway Manual Mode Overview Linux 143 v Determine VNIC Configuration Parameters For Gateway Manual Mode Linux 144 v Determine VLAN Associations for Gateway Manual Mode Linux 145 v Configure and Create VNICs for Gateway Manual Mode Linux 146 VNIC Configuration on Linux Hosts in Gateway Manual Mode 147 Creating VNICs Under Host Manual Mode Linux 148 Host Manual Mode Overview Linux 149 Central Configuration File Linux 150 VNIC Specific Configuration File Linux 151 viii Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 v v v Determine VNIC Configuration Parameters for Host Manual Mode Linux 152 Create the Central Configuration File Linux 154 Create the VNIC Specific Configuration Files Linux 156 mlx4_vnic_confd Daemon Linux 157 y Configure and Create VNICs for Host Manual Mode Linux 158 Creat
134. guration files define the number of VNICs and the VNICs properties The mlx4_vnic_confd daemon in the BXOFED software reads the configuration files and passes the relevant data to the mlx4_vnic module which creates the VNICs The two types of configuration files for VNICs provide the same functionality A central configuration file m1x4_vnic conf m VNIC specific configuration files ifcfg ethXX where XX is the eth number of the VNIC If both forms of configuration files exist the central configuration file has precedence and is the only file used When creating VNICs consider that there are conditions from partitioning See smpartition Command Guidelines on page 107 Related Information m Gateway Manual Mode Overview Linux on page 143 m Central Configuration File Linux on page 150 a VNIC Specific Configuration File Linux on page 151 m Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 m Create the Central Configuration File Linux on page 154 m Create the VNIC Specific Configuration Files Linux on page 156 m mlx4_vnic_confd Daemon Linux on page 157 m Configure and Create VNICs for Host Manual Mode Linux on page 158 Administering Gateway Resources 149 Central Configuration File Linux The central configuration file is the etc infiniband ml1x4_vnic conf file and consists of multiple single line entries each describing a
135. h use duplication disclosure modification and adaptation of the programs including any operating system integrated software any programs installed on the hardware and or documentation shall be subject to license terms and license restrictions applicable to the programs No other rights are granted to the U S Government This software or hardware is developed for general use in a variety of information management applications It is not developed or intended for use in any inherently dangerous applications including applications which may create a risk of personal injury If you use this software or hardware in dangerous applications then you shall be responsible to take all appropriate fail safe backup redundancy and other measures to ensure its safe use Oracle orporation and its affiliates disclaim any liability for any damages caused by use of this software or hardware in dangerous applications Oracle and Java are registered trademarks of Oracle and or its affiliates Other names may be trademarks of their respective owners Intel and Intel Xeon are trademarks or registered trademarks of Intel Corporation All SPARC trademarks are used under license and are trademarks or pooner trademarks of SPARC International Inc AMD Opteron the AMD logo and the AMD Opteron logo are trademarks or registered trademarks of Advanced Micro Devices UNIX is a registered trademark of The Open Group This software or hardware and documentation may p
136. he ibportstate command for external ports On the management controller type FabMan gateway_name gt ibportstate mM m_key lid port speed speed where m m_key is the management key 16 hexadecimal digits optional m lid is the LID of the node m port is the port of the node m speed is the signaling rate per lane of the port there are 4 lanes per port 1 for 2 5 Gbit sec 2 for 5 0 Gbit sec and 4 for 10 0 Gbit sec Administering the InfiniBand Fabric 103 Note Adding speed values enables either speed For example speed 7 is 2 5 5 0 and 10 0 Gbit sec For example FabMan gateway_name gt ibportstate 15 23 speed 1 Initial PortInfo Port info Lid 15 port 23 LinkSpeedEnabled 2 5 Gbps or 5 0 Gbps or 10 0 Gbps After PortInfo set Port info Lid 15 port 23 FabMan gateway_name gt ibportstate 15 23 speed 7 Initial PortInfo Port info Lid 15 port 23 LinkSpeedEnabled 4 2 5 Gbps After PortInfo set Port info Lid 15 port 23 LinkSpeedEnabled 2 5 Gbps or 5 0 Gbps or 10 0 Gbps FabMan gateway_name gt Related Information m Gateway Reference ipportstate command m Display the Link Status of a Node on page 88 V Disable or Enable an External Port If a port is found to be problematic you can disable it Note If secret M_Key functionality is enabled use the M m_key option to the ibportstate comman
137. he management controller of the gateway restricts the number of VNICs that can be created by specified host ports for specified external ports 0A ETH 1 to OA ETH 4 and 1A ETH 1 to 1A ETH 4 Using the sethostvniclimit command to allow host created VNICs is optimal because the VNIC resources are controlled by the gateway m allowhostconfig Running the allowhostconfig command on the management controller of the gateway allows all hosts connected to the gateway through the InfiniBand fabric to create unlimited VNICs on any external port OA ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 In this situation VNIC resources are not controlled by the gateway Use of the allowhostconfig command requires consideration of the impact on the gateway s performance Only one of these methods can be used at a time You can display which method is available for use with the showgwconfig command If in the output of the showgwconfig command the Running Value for the Allow host VNIC config parameter is yes then unlimited host created VNICs functionality is enabled allowhostconfig If you want to restrict host created VNICs use the disallowhostconfig command to change the Running Value to no as seen with the showgwconfig command and then use the sethostvniclimit command Related Information m Gateway Reference sethostvniclimit command m Gateway Reference allowhostconfig command m Gateway Reference showgwconfig command m Gateway Reference disal
138. he subnet If a switch or gateway having a different secret M_Key is brought into the fabric the fabrics current secret M_Key must be assigned manually 64 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Once implemented the secret M_Key functionality is transparent to the user Attempts to hijack the fabric by a rogue Subnet Manager are refused because the secret M_Key is missing or there is a secret M_Key mismatch This condition prevents the intruder from configuring the InfiniBand fabric There are prerequisites to implementing the secret M_Key functionality You must m Correctly define and distribute the smnodes list to all smnodes m For redundancy and up time configure and enable IPoIB across the fabric m Verify and propagate the fabric element configuration Configuring secret M_Key for the first time requires m Setting the replication and encryption password m Enabling M_Key persistence m Enabling secret M_Key functionality These requirements are implemented before configuring secret M_Key functionality so that the security of the fabric is maintained during the configuration Related Information m smsubnetprotection Command on page 65 m Prepare for Secret M_Key Functionality on page 66 Configuring Secret M_Keys on page 67 m Set the Replication Password on page 76 m Enable or Disable Secret M_Key Persistence on page 76 m Enab
139. health of the InfiniBand fabric 98 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 On the management controller type FabMan gateway_name gt ibdiagnet v r Related Information m Gateway Reference ibdiagnet command m Gateway Reference ibdiagpath command m Display the InfiniBand Fabric Topology Detailed on page 86 V Determine Changes to the InfiniBand Fabric Topology The generatetopology command enables you to take a snapshot of your fabric At a later time use the matchtopology command to compare the topology file to the present conditions 1 Take a snapshot of the fabric topology FabMan gateway_name gt generatetopology date topo where date is the date of the snapshot For example FabMan gateway_name gt generatetopology Sept8 topo It will take some time to generate a topology file Please wait topo conf exist will move it to topo conf old will create new topo conf Topo conf file is created Will now start generating the topo file Wrote Topology file Sept8 topo FabMan gateway_name gt 2 After an event or as part of scheduled network management compare the present topology to that saved in the topology file FabMan gateway_name gt matchtopology date topo where date is the date of the snapshot For example FabMan gateway_name gt matchtopology Sept8 topo Topology matching will take some time Please wait J
140. hernet Over InfiniBand Driver Oracle Solaris on page 162 Set Up VIOAs Oracle Solaris on page 163 Configure IP Addresses on the Data Link Oracle Solaris on page 167 Monitoring Gateway Resources These topics describe how to monitor the gateway interfaces Note To use the commands described in these topics you must access them through the SYS Gateway_Mgmt or SYS Fabric_Mgmt Linux shell targets of the Oracle ILOM CLI Display LAG Information on page 173 Display the VLANs on page 174 Display the VNICs on page 176 Display IO Adapters on page 177 Display Gateway Ethernet Port Information on page 178 Display Gateway Port Information on page 179 Display the Gateway Instance Number on page 181 172 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m Display Information About the Gateway on page 182 Related Information m Monitoring the Chassis on page 19 m Monitoring the 14 Switch Chip on page 31 m Monitoring the InfiniBand Fabric on page 83 m Monitoring the Subnet Manager on page 49 m Installing Gateway Supportive Software Linux on page 137 m Creating VNICs Under Gateway Manual Mode Linux on page 142 m Creating VNICs Under Host Manual Mode Linux on page 148 m Creating Virtual IO Adapters Oracle Solaris on page 160 m Controlling LAGs on page 183
141. hey have not been merged For these reasons you must use the force option for the smsubnetprotection command otherwise the partition daemons will prevent configuration because the nodes are unavailable 3 Determine which Subnet Manager is to be the master Subnet Manager and which are to be the standby Subnet Managers 4 Set the priority for each Subnet Manager See Set the Subnet Manager Priority on page 56 5 Physically merge the subnets into one fabric Related Information m Add a New Subnet Manager to the Secret M_Key Fabric on page 78 m Secret M_Key and Firmware Upgrade and Downgrade Guidelines on page 80 Secret M_Key and Firmware Upgrade and Downgrade Guidelines Follow these guidelines for firmware upgrades and downgrades with respect to secret M_Key functionality m For ease of use upgrade all firmware to version 2 1 or later before implementing secret M_Key m If you downgrade to firmware version 2 0 or earlier you lose secret M_Key functionality Disable secret M_Key functionality before the downgrade otherwise the remaining secret M_Key portions of the fabric become undiscoverable by the downgraded gateway or switch m This table describes the firmware type of M_Key functionality and its support by the firmware versions Functionality Firmware 1 3 and Earlier Firmware 2 0 Firmware 2 1 and Later Readable M_Key No Supported Supported Secret M_Key No No Supported Note Readable M_Keys c
142. ify a Partition or Port on page 114 m Remove User Partitions for Firmware Downgrade on page 117 V Remove User Partitions for Firmware Downgrade You must perform these steps before downgrading the firmware to a version before 2 0 Note Only perform this procedure if you are downgrading the firmware Administering the InfiniBand Fabric 117 Note By removing user partitions you might lose connectivity that was available only through those partitions Additionally you might gain undesired connectivity because all hosts become full members of the default partition 1 On the management controller of the gateways where VNICs using user partitions were created remove all VNICs that are using user partitions See Delete VNICs on page 201 2 Remove all VLANs mapped to user partitions See Delete VLANs on page 194 3 Remove all user defined partitions See Delete a Partition on page 116 4 Restore the default partition to default settings See Modify a Partition or Port on page 114 however use this command line to modify the partition FabMan gateway_name gt smpartition modify pkey 0x7fff port ALL_CAS flag ipoib m full 5 Consider your next step a If your InfiniBand fabric has just one Subnet Manager downgrade the firmware Refer to Gateway Remote Management upgrading the firmware a If your InfiniBand fabric has multiple Subnet Managers go to Step 6 6
143. ii sae rade odes eng Speers 10368 RM Was vas mse te Bee stad Sieh ang areca onde 0 FabMan gateway_name gt For example for connector 1A ETH 1 FabMan gateway_name gt getportcounters 1A ETH 1 Port counters for connector 1A ETH 1 Bridge 1 port Bridge 1 2 RX DYES so ost a Paha edi bd Aa ne Pay oS 8832 RX PaCke ts c cee edad hs Sees 69 RX Jumbo packets 0 RX unicast packets 0 RX multicast packets 69 RX broadcast packets 0 RX no b ff r r ok ede ae gw eed 0 RX CROI ees Bike Rok Ble Ne wae Bek Be we Reese 0 RX PUN es BR ae e ee A E Ree a es 0 RX Err roro sie oa eos EEG A eee a 0 TX Dy CSS oe 2s soda ee ee ees VS 8832 MK PAaCkKete veces Oi leeds a ek ea DE 69 TX Jumbo packets 0 TX unicast packets 0 TX multicast packets 69 Administering the 14 Switch Chip TX broadcast packets 0 TX SOO S nau ee Scat envied Soe ews a 0 FabMan gateway_name gt Related Information m Gateway Reference getportcounters command m Display Counters for a Node on page 89 Controlling the 14 Switch Chip These topics describe how you can manage the I4 switch chip Note To use the commands described in these topics you must access them through the SYS Fabric_Mgmt Linux shell target of the Oracle ILOM CLI m Disable a Switch Chip Port on page 40 m Enable a Switch Chip Port o
144. ing Virtual IO Adapters Oracle Solaris 160 Ethernet Over InfiniBand Overview Oracle Solaris 161 v lt a lt lt a Verify That the Ethernet Over InfiniBand Driver Is Installed Oracle Solaris 162 Install the Ethernet Over InfiniBand Driver Oracle Solaris 162 Set Up VIOAs Oracle Solaris 163 Configure IP Addresses on the Data Link Oracle Solaris 167 Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris 169 Monitoring Gateway Resources 172 v 4 lt aaa v Display LAG Information 173 Display the VLANs 174 Display the VNICs 176 Display IO Adapters 177 Display Gateway Ethernet Port Information 178 Display Gateway Port Information 179 Display the Gateway Instance Number 181 Display Information About the Gateway 182 Controlling LAGs 183 v lt a a a Enable LAG Mode 184 Create LAGs 185 Create a Global Gateway 186 Add or Delete Connectors From a LAG 187 Deletea LAG 188 Contents ix v Disable LAG Mode 189 Controlling VLANs and VNICs_ 190 VNIC Allocation on the Gateway 191 Create VLANs 192 Delete VLANs 194 Create VNICs 195 Modify VNICs 197 Enable VNICs 199 Disable VNICs 200 v Delete VNICs 201 lt a a a al Controlling Gateway Ports and Parameters 202 Enable a Gateway Ethernet Port 203 Disable a Gateway Ethernet Port 204 Set the Gateway Ethernet Port Parameters 205 Set the Gateway Service Level 206 Set the Gateway Instance Number 207 qd lt 4 lt Set the Gateway
145. ion of the smconfigtest command the Subnet Manager is expected to have a priority of 5 or 8 For the leaf option the expected priority is 5 Verify the Subnet Manager configuration FabMan gateway_name gt smconfigtest spine leaf where m spine for spine switches m leaf for leaf switches For example when the Subnet Manager has priority 6 FabMan gateway_name gt smconfigtest SubnetManager configuration is valid FabMan gateway_name gt Related Information m Gateway Reference smconfigtest command m Display Subnet Manager Status on page 50 m Display the Subnet Manager s Location History on page 51 m Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 m Display the Subnet Manager Log on page 55 54 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Vv Display the Subnet Manager Log 1 On the management controller type FabMan gateway_name gt showsmlog Jul 31 15 27 56 061000 B7D328D0 0x03 gt OpenSM 3 2 6_20120730 Oracle patch 113 OpenSM 3 2 6_20120730 Oracle patch 11 3 Reading Cached Option File etc opensm opensm conf Loading Cached Option sm_sl 1 Loading Cached Option packet_life_time 15 Sep 05 04 49 30 976000 B765AB90 0x02 gt osm_report_notice Reporting Informational Notice CapabilityMask NodeDescription Link Wi
146. irmware guidelines 80 merging subnets 79 new Subnet Manager 78 overview 64 preparing 66 setting replication password 76 setting to use 72 secure fabric management 64 service level setting 206 setgwethport command 178 205 setgwinstance command 181 207 setgwsl command 206 setgwsystemname command 208 setsmmkey command 59 60 setsmpriority command 51 56 58 setsmrouting command 61 Index 215 setsubnetprefix command 57 setting Ethernet port 205 instance number 207 port speed 103 replication password 76 secret M_Key to use 72 service level 206 Subnet Manager controlled handover 58 management key 59 prefix 57 priority 56 routing algorithm 61 system name 208 howdisk command 24 howfree command 25 howfruinfo command 26 howgwconfig command 182 howgwports command 179 howioadapters command 177 howlag command 173 howpsufru command 26 howsmlog command 55 howtemps command 22 howtopology command 86 howunhealthy command 20 howvlan command 174 Hn nnn n nnn AR AHR AHR Hn N howvnics command 176 smconfigtest command 54 sminfo command 50 smnodes adding to list 133 deleting from list 134 displaying 95 smnodes add command 133 delete command 134 list command 95 smpartition add command 112 create command 110 delete command 116 list command 93 modify command 114 remove command 112 start command 110 smpartition command 106 guidelines 107 smpquery comman
147. istence on page 76 m Enable or Disable Secret M_Key Functionality on page 77 m Considering Situations In Regard to Secret M_Key on page 78 Related Information m Monitoring the Subnet Manager on page 49 m Controlling the Subnet Manager on page 56 Secret M_Key Overview The secret M Key feature of the version 2 1 firmware adds an extra layer of security to the InfiniBand fabric through use of a hidden passphrase or secret M_Key The secret M_Key enforces a policy that only a Subnet Manager authorized to manage the InfiniBand fabric is able to do so and that write requests from unauthorized sources are rejected Conveniently information can be read from the InfiniBand devices without restriction except for the secret M_Key itself In the process of initiating the secret M_Key functionality you create the secret M_Key on the master Subnet Manager using the smsubnetprotection command and its subcommands in a configuration session During the session one or more secret M_Keys are added to a configuration list There is a maximum of 10 secret M_Keys Upon committing to the configuration the list and active secret M_Key is automatically propagated to the standby Subnet Managers in the InfiniBand fabric This action authorizes and enables the standby Subnet Managers to control the fabric in the event of a failover or handover The master Subnet Manager also sets the value of the secret M_Key to the InfiniBand nodes of t
148. ithm 51 status 50 topology detailed 86 simple 86 VLANs 174 VNICs 176 dladm command 163 167 169 E element adding 129 modifying 131 removing 129 enablegwport command 203 enablelagmode command 184 enablesm command 62 enableswitchport command 41 45 enablevnic command 199 enabling gateway Ethernet port 203 LAG mode 184 secret M_Key functionality 77 persistence 76 Subnet Manager 62 switch chip port 41 VNICs 199 env_test command 23 EoIB 161 installing driver 162 verifying installation 162 error counters clearing 102 displaying 89 Ethernet displaying information 178 setting port 205 example route through gateway 9 F fabric diagnostics 98 displaying configuration information 93 element 125 Fabric Director 125 fabric elements 119 adding elements to configuration 122 configurations 120 creating configuration 120 deleting elements from configuration 122 displaying 96 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 modifying elements of configuration 123 fan checking speed 22 status 22 fdconfig abort command 127 complete fabric config command 127 define element command 129 define fabric name command 127 list current fabric config command 94 redefine element command 131 remove element command 129 start fabric config empty command 127 fdconfig command 126 firmware downgrades and partiti
149. ition_name pkey p_key flag ipoib mtu mtu rate rate sl sl scope scope m defmember where m partition_name is an alphanumeric tag to the InfiniBand partition optional m p_key is the partition key 1 to 7 or default a mtu is the number that maps to the actual MTU 1 to 5 mtu Number 1 2 3 4 5 MTU Value 256 512 1024 2048 4096 m rate is the number that maps to the actual throughput of a link link width link speed 2 to 10 110 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 rate Number 2 3 4 5 6 7 8 9 10 Rate Value in Gbps 2 5 10 30 5 20 40 60 80 120 m sl is the service level 0 to 15 Note Use service level 1 s1 1 only for low latency high priority small message low bandwidth traffic Use other service levels for regular high bandwidth traffic m scope is the multicast address scope value 1 to 14 Note The mtu rate s1 and scope parameters are for the multicast group created when ipoib IP over InfiniBand is configured for the partition Typically these values are not specified as the defaults are sufficient for the fabric configuration m defmember is the default membership type full limited or both for the partition Note If ports are added to the partition without specifying the membership type the default membership type is applied to the port For example FabMan gateway
150. lated Information a Gateway Hardware Problems on page 1 a Network Management Troubleshooting Guidelines on page 6 m Understanding Routing Through the Gateway on page 7 a Switch GUIDs Overview on page 10 Network Management Troubleshooting Guidelines Follow these guidelines when experiencing difficulty connecting to the gateway through its NET MGT port m Ensure that the state property of the SP network Oracle ILOM target is set to enabled For example gt set SP network state enabled m Verify the integrity of the physical connection between the NET MGT port of the server the network and the management device m Ensure that at least one of the appropriate network services are enabled in Oracle ILOM Service Command HTTP set SP services http secureredirect disabled servicestate enabled HTTPS set SP services https servicestate enabled m Use an industry standard network diagnostic tool like IPv4 Ping or Traceroute to test the network connections between the server and the management device 6 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information m Gateway Hardware Problems on page 1 a InfiniBand Fabric Problems on page 3 m Understanding Routing Through the Gateway on page 7 m Switch GUIDs Overview on page 10 Understanding Routing Through the Gateway The tables in these topics descri
151. le FabMan gateway_name gt ibtracert 15 14 From switch 0x00212856cd22c0a0 portnum 0 lid 15 15 SUN IB QDR GW switch mnm34 97 gt ca port 0x00212856cd22c042 2 lid 14 14 SUN IB QDR GW switch mnm34 97 Bridge 1 To ca 0x00212856cd22c040 portnum 2 lid 14 14 SUN IB QDR GW switch mnm34 97 Bridge 1 FabMan gateway_name gt 88 Note The output for your InfiniBand fabric will differ from that in the example For this example m The route starts at the switch with GUID 0x00212856cd22c0a0 and is using port 0 The switch is LID 15 and in the description the switch host s name is mnm34 97 m The route enters at port 1 of the CA with GUID 0x00212856cd22c042 and exits at port 2 The CA is LID 14 and is within bridge chip 1 Related Information m Gateway Reference ibtracert command Vv Display the Link Status of a Node If you want to know the link status of a node in the InfiniBand fabric the ibportstate command can tell you the state width and speed of that node On the management controller type FabMan gateway_name gt ibportstate lid port where m lid is the LID of the node Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m port is the port of the node For example FabMan gateway_name gt ibportstate 15 23 PortiInfo Port info Lid 15 port 23 bPnkStabe s eado 80d Saleen ee eae oc Active PhysLinkState
152. le or Disable Secret M_Key Functionality on page 77 m Considering Situations In Regard to Secret M_Key on page 78 smsubnetprotection Command You can enable secret M_Key functionality using the smsubnetprotection command and its subcommands The smsubnetprotection command is primarily issued from the gateway or switch that has the master Subnet Manager and acts upon the secret M_Key configuration There are two configurations the active configuration is the one currently in use and the modified configuration is the one which you can affect Configuring the secret M_Key consists of starting a configuration session adding or deleting M_Keys from the list setting the current M_Key and committing the list to active implementation During the configuration session you can also view the configuration list undo your last action or abort the configuration session altogether Administering the Subnet Manager 65 If you will only add or only delete secret M_Keys from the list during the configuration session you must initiate the session with the smsubnetprotection start addonly or smsubnetprotection start deleteonly command Note You cannot both add and delete secret M_Keys within a single configuration session You must perform these actions in separate configuration sessions The smsubnetprotection command can also enable and disable secret M_Key functionality display the active M_Keys set and clear the local secret M
153. ller type FabMan gateway_name gt deletelag lagname where lagname is the identifier of the LAG For example FabMan gateway_name gt deletelag 0A12 LAG OA12 deleted FabMan gateway_name gt Related Information m Gateway Reference deletelag command a Enable LAG Mode on page 184 m Create LAGs on page 185 m Add or Delete Connectors From a LAG on page 187 m Disable LAG Mode on page 189 Disable LAG Mode When you disable LAG mode the LAG configurations remain Upon re enabling LAG mode the LAGs become active again Note LAG mode must be disabled for all InfiniBand gateways in the fabric 1 On the management controller type FabMan gateway_name gt disablelagmode Stopping Bridge Manager OK Starting Bridge Manager OK FabMan gateway_name gt 2 Repeat Step 1 for all InfiniBand gateways in the fabric Related Information m Gateway Reference disablelagmode command m Enable LAG Mode on page 184 m Create LAGs on page 185 m Add or Delete Connectors From a LAG on page 187 Administering Gateway Resources 189 m Delete a LAG on page 188 Controlling VLANs and VNICs These topics describe how you can control the gateway ports and parameters Note To use the commands described in these topics you must access them through the SYS Gateway_Mgmt or SYS Fabric_Mgmt Linux shell targets of the Oracle ILOM CLI a VNIC Alloc
154. lowhostconfig command m Create VNICs on page 195 m Enable VNICs on page 199 m Disable VNICs on page 200 m Delete VNICs on page 201 Administering Gateway Resources 191 V Create VLANs When you create a VLAN you create a mapping between a gateway connector a VLAN identifier and a partition key You can create VLANs with the createvlan command If a VNIC will be associated to a VLAN you must create the VLAN before the VNIC Note Due to hardware limitations for MultiCast groups there is a maximum of 1000 VLANs Note Regardless if you plan to use VLANs you must configure a no VLAN to P_Key relationship For example the VLAN ID No is assigned to the P_Key default Note The association of VLAN identifier NO to partition key default survives firmware upgrades 1 If unknown determine the connectors VNICs and partition keys to be associated with the VLANs See Determine VLAN Associations for Gateway Manual Mode Linux on page 145 2 Optional Use the gateway MAC address list that you created earlier See Determine VLAN Associations for Gateway Manual Mode Linux on page 145 3 Create the VLAN on the management controller FabMan gateway_name gt createvlan connector lagname vlan vlan_ID pkey p_key where m connector is the name of the connector OA ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH This value is column one of the
155. m Partitioning the InfiniBand Fabric on page 105 Administering the InfiniBand Fabric Correlating Fabric Elements on page 119 Configuring the Fabric Director Node List on page 125 Performing Supportive Tasks on page 132 Related Information Controlling the Chassis on page 28 Controlling the 14 Switch Chip on page 40 Controlling Gateway Ports and Parameters on page 202 Controlling the Subnet Manager on page 56 Controlling Gateway Ports and Parameters on page 202 Monitoring the InfiniBand Fabric on page 83 Investigating Nodes These tasks help you determine the condition of the InfiniBand fabric Perform Comprehensive Diagnostics for the Entire Fabric on page 98 Determine Changes to the InfiniBand Fabric Topology on page 99 Find 1x SDR or DDR Links in the Fabric on page 100 Determine Which Links Are Experiencing Significant Errors on page 100 Related Information Controlling Nodes on page 101 Partitioning the InfiniBand Fabric on page 105 Correlating Fabric Elements on page 119 Configuring the Fabric Director Node List on page 125 Performing Supportive Tasks on page 132 Vv Perform Comprehensive Diagnostics for the Entire Fabric If you require a full testing of your InfiniBand fabric the ibdiagnet command can perform many tests with verbose results The command is a useful tool to determine the general overall
156. m Gateway Remote Management understanding Oracle ILOM commands m Hardware Command Overview on page 16 a InfiniBand Command Overview on page 17 Hardware Command Overview The management controller uses a simplified Linux OS and file system From the FabMan gateway_name gt prompt on the management controller you can type hardware commands to perform administrative and management tasks and configure partitions the InfiniBand fabric the Subnet Manager VNICs and VLANs After you log in to the ilom admin account use the show SYS Fabric_Mgmt command to obtain the restricted Linux shell prompt FabMan gateway_name gt You can now enter hardware commands in this format FabMan gateway_name gt command arguments arguments 16 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information Gateway Reference understanding hardware commands m Oracle ILOM Command Overview on page 16 a InfiniBand Command Overview on page 17 InfiniBand Command Overview The InfiniBand commands are a means of monitoring and controlling aspects of the InfiniBand fabric These commands are also installed on and run from the management controller which is also the host of the Subnet Manager Use of these commands requires thorough knowledge of InfiniBand architecture and technology After you log in to the ilom admin account use the show SYS Fabric_Mgmt command to ob
157. move smpartition modify and smpartition delete commands You finally make the modified configuration into the active configuration with the smpartition commit command Refer to Gateway Reference smpartition command for more information Note There is a hardware limitation of 127 user defined and 1 default partition for each port of the ConnectX2 chips in HCAs and for each InfiniBand port of the BridgeX chips in gateways Related Information m Gateway Reference smpartition command smpartition Command Guidelines on page 107 m Partitions and P_Keys on page 108 m Determine the Partitions and P_Keys on page 109 m Create a User Partition on page 110 m Add or Remove a Port From a Partition on page 112 a Modify a Partition or Port on page 114 106 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Delete a Partition on page 116 Remove User Partitions for Firmware Downgrade on page 117 smpartition Command Guidelines When partitioning the InfiniBand fabric with the smpartition command consider these guidelines during the configuration session Valid P_Keys for partitioning are 0x1 through 0x7fff However both 0x1 and 0x7fff are predefined P_Keys and cannot be created or deleted with the smpartition command Default entries ports in the predefined partitions cannot be removed However when you add ports into the p
158. n If you have partitioned the InfiniBand fabric you can display the active in use partition configuration or the modified in standby partition configuration with the smpartition command 1 On the management controller type FabMan gateway_name gt smpartition list configuration where configuration is the partition configuration name active or modified For example FabMan gateway_name gt smpartition list active Sun DCS IB partition config file version_number 1 version_number 12 Default 0x7fff ipoib ALL_CAS full ALL_SWITCHES full Administering the InfiniBand Fabric 93 SELF full SUN_DCS 0x0001 ipoib ALL_SWITCHES full 0x0004 ipoib 0x0021280001cf 3787 full 0x0021280001cf 205b full FabMan gateway_name gt Note The output is displayed one page at a time In this short example in addition to the default partition Default 0x7fff there is a new partition with P_Key 0x0004 This partition is configured for IPoIB and has two member ports that are assigned full membership of the partition 2 To advance to the next page of output press the spacebar Repeat as necessary 3 If the information displayed is incorrect modify the partition configuration See Modify a Partition or Port on page 114 Related Information m Gateway Reference smpartition command m Partitioning the InfiniBand Fabric on page 105 m Di
159. n 0x0006 Firmware Revision 0x0102 Administering the Chassis BIOS version NOW1R112 BIOS date 04 24 2009 FabMan gateway_name gt Related Information m Gateway Reference version command m Gateway Remote Management version command Oracle Auto Service Request Oracle Auto Service Request is a secure customer installable support feature that automatically generates a service request for specific hardware faults The software available as a feature of the Oracle hardware warranty and Oracle Premier Support for Systems can improve system availability through expedited diagnostics and priority service request handling Please contact Oracle support for details on setting up ASR on your gateway Related Information a Troubleshooting the Gateway on page 1 m Gateway Service detecting and managing faults Controlling the Chassis These topics describe how you can manage the gateway and its components Note To use the commands described in these topics you must access them through the SYS Fabric_Mgmt Linux shell target of the Oracle ILOM CLI m Restart the Management Controller on page 29 m Change the root User Password on page 29 Related Information m Controlling the I4 Switch Chip on page 40 Controlling Gateway Ports and Parameters on page 202 m Controlling the InfiniBand Fabric on page 97 28 Sun Network QDR InfiniBand Gateway Switch Administration Gui
160. n files and fabric MIBs are created Related Information m Gateway Reference dconfig command m Display the InfiniBand Fabric Configuration on page 94 m Fabric Director and Fabric Elements on page 125 m fdconfig Command Overview on page 126 m Add or Remove an Element From the Fabric Configuration on page 129 m Modify an Element of the Fabric Configuration on page 131 V Add or Remove an Element From the Fabric Configuration 1 Initiate a fabric configuration session FabMan gateway_name gt fdconfig start fabric config from current FabMan gateway_name gt 2 Add an element to the fabric configuration FabMan gateway_name gt dconfig define element name name ip I P_address type type role rolel role role2 where m name is the host name of the element as configured by Oracle ILOM m P_address is the IP address of an element m type is the identifier of the type of element sw36 Sun Datacenter InfiniBand Switch 36 a sw36gw Sun Network QDR InfiniBand Gateway Switch a unknown The element type is unknown m role is the purpose of the element a gateway The element acts as a gateway spineSwitch The element acts as a spine switch a leafSwitch The element acts as a leaf switch Administering the InfiniBand Fabric 129 unknown The element role is unknown For example FabMan gateway_name gt fdconfig define element name secondary ip 123 45
161. n page 112 m Delete a Partition on page 116 m Remove User Partitions for Firmware Downgrade on page 117 Delete a Partition When you delete a partition you effectively commit a blank default configuration 1 Initiate a partition configuration session on the management controller FabMan gateway_name gt smpartition start FabMan gateway_name gt Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 2 Delete the partition FabMan gateway_name gt smpartition delete n partition_name pkey p_key where m partition_name is an alphanumeric tag to the InfiniBand partition optional m p_key is the partition key 2 to 7ffe Note You cannot delete the pre defined partitions with P_Keys 1 and 7fff For example FabMan gateway_name gt smpartition delete n testpartition FabMan gateway_name gt 3 End the configuration session and commit the configuration to the active partition configuration FabMan gateway_name gt smpartition commit FabMan gateway_name gt Related Information m Gateway Reference smpartition command m smpartition Command on page 106 a smpartition Command Guidelines on page 107 m Partitions and P_Keys on page 108 m Determine the Partitions and P_Keys on page 109 m Create a User Partition on page 110 m Add or Remove a Port From a Partition on page 112 m Mod
162. n page 41 a Autodisabling Ports and Connectors on page 43 Related Information a Controlling the Chassis on page 28 a Controlling Gateway Ports and Parameters on page 202 m Controlling the InfiniBand Fabric on page 97 m Controlling the Subnet Manager on page 56 m Monitoring the 14 Switch Chip on page 31 Vv Disable a Switch Chip Port On occasion you might need to turn off a port For example a cable might become damaged and cause symbol errors that affect the switch chip s performance Use the disableswitchport command to disable a switch chip port and its respective cable connection Note The disableswitchport command is persistent and survives reboots 40 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 On the management controller type FabMan gateway_name gt disableswitchport reason reason Switch port where m reason is the reason for disabling the port Blacklist or Partition m portis the number of the port 1 to 36 Alternatively type FabMan gateway_name gt disableswitchport reason reason connector where connector is the number of the QSFP connector 0A to 15B For example FabMan gateway_name gt disableswitchport Switch 14 Disable Switch port 14 Initial PortInfo Port info DR path slid 65535 dlid 65535 0 port 14 LinkState eeg ae esua E goles E oes Active PhysSLinkStat
163. nabled Bridge 1 Port 1A ETH 3 Bridge 1 1 up Enabled Bridge 1 Port 1A ETH 4 Bridge 1 1 up Enabled Connector 9B Present lt gt Switch Port 13 up Enabled FabMan gateway_name gt If connector and its link have been autodisabled the 1istlinkup command will append the connector status with AutomaticHighErrorRate AutomaticBadSpeedOrwWidth or both For example Connector OA Present lt gt Switch Port 20 down AutomaticHighErrorRate In this situation check the condition of the connector and cable Refer to Gateway Service servicing cables Related Information m Gateway Reference listlinkup command m Display the Link Status of a Node on page 88 Vv Display Connectors Configured With Autodisable Functionality You can display the connectors that are configured to automatically disable in the presence of high error rates or suboptimal link speed or width 36 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 On the management controller type FabMan gateway_name gt autodisable list Connectors which will be disabled on high error rate OA OB 1A 1B 2A 2B 3A 3B 4A 4B 5A 5B 6A 6B 7A 7B 8A 8B 9A 9B 10A 10B 11A 11B 12A 12B 13A 13B 14A 14B 15A 15B Connectors which will be disabled on suboptimal link speed or width OA OB 1A 1B 2A 2B 3A 3B 4A 4B 5A 5B 6A 6B 7A 7B 8A 8B 9A 9B 10A 10B 11A 11B 12A 12B 13A 13B 14A 14B 1
164. nagement controller type FabMan gateway_name gt checkboot Switch OK Bridge 0 OK Bridge 1 OK All Internal ibdevices OK FabMan gateway_name gt Related Information m Gateway Reference checkboot command a Display Gateway Environmental and Operational Data on page 23 m Gateway Reference ipbdevreset command Vv Display Link Status In some situations you might need to know the status of each route through the gateway Additionally the 1istlinkup command displays where InfiniBand cables are connected to the gateway Administering the 14 Switch Chip 35 On the management controller type FabMan gateway_name gt listlinkup connected Connector 2A Present lt gt Switch Port 24 up Enabled Connector 3A Present lt gt Switch Port 26 up Enabled Connector 6A Present lt gt Switch Port 35 up Enabled Connector 8A Present lt gt Switch Port 31 up Enabled Connector 11A Present lt gt Switch Port 12 down Enabled Connector 13A Present lt gt Switch Port 9 down Enabled Connector 14A Present lt gt Switch Port 7 down Enabled Connector OA ETH Present Ee Aa ae E Bridge 0 Port OA ETH 1 Bridge 0 2 up Enabled Bridge 0 Port OA ETH 2 Bridge 0 2 up Enabled Bridge 0 Port OA ETH 3 Bridge 0 1 up Enabled Bridge 0 Port OA ETH 4 Bridge 0 1 up Enabled Connector 1A ETH Present Bridge 1 Port 1A ETH 1 Bridge 1 2 up Enabled Bridge 1 Port 1A ETH 2 Bridge 1 2 up E
165. nector is the name of the connector 0A to 15A and OB to 15B m cause is the reason to disable the connector a H The error rate is higher than what the InfiniBand specification permits a L The link speed or width is less than optimal less than 4x QDR Alternatively to add all connectors to the autodisable list with both causes type FabMan gateway_name gt autodisable all For example to add connector 0A to the autodisable list for high error rates FabMan gateway_name gt autodisable add 0A H FabMan gateway_name gt Related Information m Gateway Reference autodisable command m Autodisable Functionality on page 43 m Reenable Autodisabled Connectors on page 45 m Delete Connectors From the Autodisable List on page 46 V Reenable Autodisabled Connectors 1 Investigate why the connector was autodisabled Inspect the condition of the cable and connectors at both ends and replace if necessary Refer to Gateway Service servicing cables 2 On the management controller type FabMan gateway_name gt enableswitchport automatic port connector where m port is the number of the port 5 to 36 Administering the 14 Switch Chip 45 connector is the number of the QSFP connector 0A to 15B For example FabMan gateway_name gt enableswitchport automatic 0A 3 If you did not replace the cable in Step 1 you must retrain the links for the por
166. nectors to the Autodisable List on page 44 m Reenable Autodisabled Connectors on page 45 Administering the 14 Switch Chip 47 48 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Administering the Subnet Manager These topics describe the administration of the Subnet Manager m Monitoring the Subnet Manager on page 49 m Controlling the Subnet Manager on page 56 m Configuring Secure Fabric Management on page 64 Related Information m Troubleshooting the Gateway on page 1 m Understanding Administrative Commands on page 13 m Administering the Chassis on page 19 m Administering the 14 Switch Chip on page 31 m Administering Gateway Resources on page 137 m Administering the InfiniBand Fabric on page 83 Monitoring the Subnet Manager These topics describe how to monitor the Subnet Manager Note To use all the commands described in these topics you must access them through the SYS Fabric_Mgmt Linux shell target of the Oracle ILOM CLI m Display Subnet Manager Status on page 50 m Display the Subnet Manager s Location History on page 51 a Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 m Display the Secret M_Key Configuration on page 52 m Verify M_Key Persistence on page 53 m Verify the Subnet Manager Co
167. nectors used for your VNICs you can find the respective BridgeX chip s InfiniBand port GUIDs to add to the non default partition Related Information m smpartition Command on page 106 m Partitions and P_Keys on page 108 m Determine the Partitions and P_Keys on page 109 a Create a User Partition on page 110 m Add or Remove a Port From a Partition on page 112 m Modify a Partition or Port on page 114 m Delete a Partition on page 116 m Remove User Partitions for Firmware Downgrade on page 117 Partitions and P_Keys The gateway supports several partitions in the InfiniBand fabric You assign a P_Key to each partition as a means of identifying the partition to the Subnet Manager P_Keys are 15 bit integers and have a value of 0x1 to 0x7fff The P_Key value is used by the master Subnet Manager to configure the InfiniBand fabric A P_Key value of 0x7fff represents the default partition An additional bit the membership bit can identify the membership of the partition m Full The membership bit is 1 Full membership permits communication to all members within a partition a Limited The membership bit is 0 Limited membership permits communication only with a full member Combined together P_Key and the membership bit comprise a 16 bit integer and the most significant bit is the membership bit In a full membership the membership bit is set high When this happens the P_Ke
168. nfiguration Integrity on page 54 49 Display the Subnet Manager Log on page 55 Related Information Monitoring the Chassis on page 19 Monitoring the 14 Switch Chip on page 31 Monitoring Gateway Resources on page 172 Monitoring the InfiniBand Fabric on page 83 Controlling the Subnet Manager on page 56 Configuring Secure Fabric Management on page 64 Vv Display Subnet Manager Status If you want to quickly determine the master Subnet Manager s state the getmaster command provides the LID and GUID of the InfiniBand device hosting the master Subnet Manager On the management controller type FabMan gateway_name gt getmaster Local SM enabled and running state MASTER 20120831 09 45 17 Master SubnetManager on sm lid 1 sm guid 0x3ba01234567a0a0 SUN DCS 36P QDR mnm2 10 172 144 66 FabMan gateway_name gt Related Information Gateway Reference getmaster command Display the Subnet Manager s Location History on page 51 Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 Verify the Subnet Manager Configuration Integrity on page 54 Display the Subnet Manager Log on page 55 50 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Vv Display the Subnet Manager s Location History On the management controller type FabMan ga
169. nfiniBand fabric Note Because of its silent nature have redundancy for autodisable configured links You can monitor autodisable configured connectors and their links using the listlinkup command or the Fabric Monitor For both interfaces a connector which has been autodisabled is identified with the text AutomaticHighErrorRate AutomaticBadSpeedOrwWidth or both Removing a connector from the autodisable list stops autodisable functionality for that connector and its link However it does not re enable an autodisabled connector or its link You must investigate the condition of the disabled cable and connector and replace it if necessary You can then re enable an autodisabled connector with the automatic option of the enableswitchport command Related Information m Gateway Reference autodisable command m Gateway Reference enableswitchport command a Add Connectors to the Autodisable List on page 44 m Reenable Autodisabled Connectors on page 45 m Delete Connectors From the Autodisable List on page 46 m Gateway Service servicing cables V Add Connectors to the Autodisable List You add connectors to either or both parts of the autodisable list with the autodisable command 44 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 On the management controller type FabMan gateway_name gt autodisable add connector cause where m con
170. nistration Guide for Firmware Version 2 1 e March 2013 Related Information m Gateway Reference dcsport command m Understanding Routing Through the Gateway on page 7 V Locate a InfiniBand Device and Connector From the GUID and Port The dcsport command can also identify the connector of an InfiniBand device that maps to a GUID and port On the management controller type FabMan gateway_name gt desport guid guid port port where m guid is the global unique identifier m port is the number of the port 1 to 36 For example FabMan gateway_name gt dcsport guid 0x00212856d102c0a0 port 36 DCS GW Switch port 36 maps to connector 6B FabMan gateway_name gt Related Information m Switch GUIDs Overview on page 10 m Gateway Reference dcsport command Vv Locate a Switch Chip or Connector From the GUID and Port The output of some InfiniBand commands identify a node by its GUID You can determine the switch chip or InfiniBand cable connection associated with that GUID using this procedure Administering the 14 Switch Chip 33 1 In the output of the InfiniBand command identify the GUID and port in question For example given this output from the ibdiagnet command W 1id 0x000f guid 0x002128312345c0a0 dev 48438 Port 21 In this example the GUID is 0x002128312345c0a0 and the port is 21 2 Use the ibswitches command to verify the GUID is that of a switch For example Fab
171. nt controller set the maximum number of host created VNICs Host Manual Mode per host port per gateway connector FabMan gateway_name gt sethostvniclimit connector host hostname port port limit number vlanlist vlan_ID1 vlan_IDn where connector is the name of the connector OA ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 m hostname is the host name of the management controller or system name recognized by the BridgeX manager m port is the port of the host m number is the maximum number of VNICs for the IOA m vlan_ID is the VLAN identifier NO or 2 to 4094 158 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN For example to enable Host Manual Mode for the host cupcake and limit it to 4 VNICs FabMan gateway_name gt sethostvniclimit OA ETH 1 host cupcake port 1 limit 4 vlanlist 3 FabMan gateway_name gt 2 Repeat Step 1 for all hosts in the fabric that need to be enabled and regulated 3 Determine if you will use a central configuration file or VNIC specific configuration files See m Central Configuration File Linux on page 150 a VNIC Specific Configuration File Linux on page 151 4 Determine the VNIC configuration
172. o configuration 129 fabric elements to configuration 122 ports to partition configuration 112 secret M_Keys 69 smnodes to list 133 addlagport command 187 administering chassis 19 gateway resources 137 hosts 137 14 switch chip 31 InfiniBand fabric 83 Subnet Manager 49 administrative command overview 13 ASR 28 autodisable command 36 44 46 autodisabled connectors adding 44 deleting 46 displaying configured 36 overview 43 reenabling 45 autodisabling ports and connectors 43 automatic service request 28 BXOFED acquiring 138 installing 137 139 Cc central configuration file 150 changing password root user 29 chassis administering 19 controlling 28 FRU ID 26 monitoring 19 checkboot command 34 checking board temperature 22 voltage 21 chassis status 23 environmental data 23 fan speed 22 status 22 flash drive status 24 link status 35 port counters 38 port status 37 power supply 20 switch chip boot status 34 checkpower command 20 checkvoltages command 21 clear data counters 102 error counters 102 management key 60 command addlagport 187 autodisable 36 44 46 checkboot 34 checkpower 20 checkvoltages 21 209 210 create_ipoib 96 135 createfabric 120 add element 122 delete element 122 list active 96 modify element 123 start 120 createlag 185 186 createvlan 192 createvnic 195 dcsport 32 delete_ipoib 136 deletelag 188 deletevlan 194 del
173. oblem replace that component or the gateway Refer to Gateway Service servicing the data cables Refer to the remote port s documentation for replacement procedures Refer to Gateway Service removing the gateway from the rack Refer to Gateway Installation installing the gateway into the rack 1 Clear the error counters See Clear Data and Error Counters on page 102 2 Start a fabric stress test 3 Identify the suspect links using the ibdiagnet command See Determine Which Links Are Experiencing Significant Errors on page 100 Look for text like this W 11id 0x0006 guid 0x0021283a8816c0a0 dev 48438 Port 34 Performance Monitor counter Value link_recovery_error_counter 0x1 symbol_error_counter 0x25 Increase by 3 during ibdiagnet 4 For links that are experiencing recovery errors or substantial symbol errors refer to other parts of this table to help identify the cause and rectify the problem Troubleshooting the Gateway 5 Situation Corrective Steps Output of 1 You can find the location of a node in the gateway by deconstructing the node s GUID InfiniBand and port commands provides See Locate a Switch Chip or Connector From the GUID and Port on page 33 only GUID and port not switch chip or QSFP connectors 2 Use the dcsport command to provide port to connector and connector to port mapping See Display the Switch Chip Port to QSFP Connector Mapping on page 32 Re
174. ode or host 9 Determine your next steps If no port GUID of the Oracle Solaris 11 host is listed or there are no MAC addresses assigned to the port GUID go to Step 10 Otherwise go to Configure IP Addresses on the Data Link Oracle Solaris on page 167 10 Determine your MAC addresses and whether they are to be global or locally administrated Note Each MAC address must be unique and not 00 00 00 00 00 00 Note Only even numbers are supported for the most significant byte of the MAC address unicast 11 Activate the VIOA by assigning a MAC address to it FabMan gateway_name gt createvnic connector guid guid mac mac_address pkey default where Administering Gateway Resources 165 connector is the name of the connector O0A ETH 1 to 0A ETH 4 and 1A ETH 1 to 1A ETH 4 m guid is the global unique identifier of the target port on the host associated with the VIOA m mac_address is the MAC address to be assigned to the VIOA Note You can use the information from Step 8 as a guide For example to assign the MAC address of a0 a5 91 95 30 9a to the VIOA for port GUID 21280001A0A591 through connector 0A ETH 2 FabMan gateway_name gt createvnic OA ETH 2 guid 0021280001A0A591 mac a0 a5 91 95 30 9a pkey default VNIC created FabMan gateway_name gt The VIOA is created and an Ethernet over InfiniBand data link is bound to the VIOA 12 Verify that the V
175. om pls topic lookup ctx acc amp id info or http www oracle com pls topic lookup ctx acc amp id trs Visit if you are hearing impaired xii Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Troubleshooting the Gateway These topics help you resolve some basic problems that might occur with your gateway m Gateway Hardware Problems on page 1 m InfiniBand Fabric Problems on page 3 m Network Management Troubleshooting Guidelines on page 6 m Understanding Routing Through the Gateway on page 7 m Switch GUIDs Overview on page 10 Related Information m Understanding Administrative Commands on page 13 m Administering the Chassis on page 19 m Administering the 14 Switch Chip on page 31 m Administering Gateway Resources on page 137 m Administering the InfiniBand Fabric on page 83 m Administering the Subnet Manager on page 49 Gateway Hardware Problems This table lists situations that might occur with gateway hardware and corrective steps that you can take to resolve the problem Situation Corrective Steps The Attention LED ona 1 Check the power supply status power supply is litorthe See Display Power Supply Status on page 20 power supply seems 2 Unplug the respective power cord wait 15 minutes then reattach the power cord dysfunctional 3 If the previous steps do not rectify the situati
176. on replace the power supply Refer to Gateway Service servicing power supplies Situation Corrective Steps There is no network 1 communication with the management controller 5 The Attention LED on 1 the gateway chassis is lit or the management controller seems 2 dysfunctional 3 4 5 The Attention LED ona 1 fan is lit or the fan seems dysfunctional 2 Verify that the management controller is powered without faults Refer to Gateway Service checking the chassis status LEDs Verify that there is a LinkUp condition at the NET MGT port Refer to Gateway Service checking the NET MGT port LEDs Verify that the DHCP server is providing the IP address you are using to access the management controller See DHCP server documentation If you are able to access the management controller through the USB connector restart the management controller Refer to Gateway Installation accessing the management controller and Restart the Management Controller on page 29 If you are unable to access the management controller through the USB connector power cycle the gateway Refer to Gateway Service powering off power supplies and Gateway Service powering on power supplies If the previous steps do not rectify the situation replace the gateway Refer to Gateway Service removing the gateway from the rack Refer to Gateway Installation installing the gateway into the rack If you are
177. on page 65 Configuring Secret M_Keys on page 67 Set the Replication Password on page 76 Enable or Disable Secret M_Key Persistence on page 76 Enable or Disable Secret M_Key Functionality on page 77 Considering Situations In Regard to Secret M_Key on page 78 Configuring Secret M_Keys Use these topics to create and administer the secret M_Key configuration Create the Secret M_Key Configuration on page 67 Add Secret M_Keys to the Configuration on page 69 Set the Secret M_Key to Use on page 72 Delete Secret M_Keys From the Configuration on page 73 Configure a Local Secret M_Key on page 75 Related Information Secret M_Key Overview on page 64 smsubnetprotection Command on page 65 Prepare for Secret M_Key Functionality on page 66 Set the Replication Password on page 76 Enable or Disable Secret M_Key Persistence on page 76 Enable or Disable Secret M_Key Functionality on page 77 Considering Situations In Regard to Secret M_Key on page 78 Vv Create the Secret M_Key Configuration 1 Prepare for secret M_Key functionality See Prepare for Secret M_Key Functionality on page 66 Administering the Subnet Manager 67 2 Initiate a secret M_Key configuration session on the management controller that hosts the master Subnet Manager FabMan gateway_name gt smsubnetprotection start addonly FabMan gateway_name gt
178. ons 117 gateway 27 secret M_Key guidelines 80 FRU ID chassis 26 power supply 26 full membership 109 G gateway controlling parameters 202 ports 202 displaying information 179 182 Ethernet port disabling 204 enabling 203 manual mode 142 configuration parameters 144 overview 143 VLAN associations 145 VNIC conditions 147 VNIC creation 146 resources administering 137 monitoring 172 troubleshooting 1 generatetopology command 99 getfanspeed command 22 getmaster command 51 getportcounters command 38 getportstatus command 37 Global Gateway 186 GUID mapping to CXP connector 33 to switch chip 33 nomenclature 10 guidelines firmware secret M_Key 80 smpartition 107 H hardware commands overview 16 hardware problems 1 HCA identity 85 host administering 137 manual mode 148 central configuration file 150 configuration parameters 152 overview 149 VNIC creation 158 VNIC specific configuration file 151 l I4 switch chip administering 31 controlling 40 monitoring 31 ibdiagnet command 98 100 102 ibhosts command 85 ibnetdiscover command 86 ibportstate command 88 102 103 104 ibswitches command 84 ibtracert command 87 identifying HCAs 85 switches 84 InfiniBand commands 17 creating configuration 127 displaying configuration 94 Index 213 fabric administering 83 controlling 97 monitoring 83 problems 3 installing BXOFED software 137 139 EoI
179. ort on page 114 m Delete a Partition on page 116 m Remove User Partitions for Firmware Downgrade on page 117 V Add or Remove a Port From a Partition You can add or remove one or more ports from the partition at one time 1 Initiate a partition configuration session on the management controller FabMan gateway_name gt smpartition start FabMan gateway_name gt 112 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 2 Add ports to the partition FabMan eateway_name gt smpartition add n partition_name pkey p_key port port ALL_CAS ALL SWITCHES ALL ROUTERS m member where m partition_name is an alphanumeric tag to the InfiniBand partition optional m p_key is the partition key 1 to 7fff m port is the GUID of the port or the special parameter to add ALL CAS All CAs in the InfiniBand fabric ALL_SWITCHES All switches ALL ROUTERS All routers m member is the membership type full limited or both for the port For example FabMan gateway_name gt smpartition add n testpartition port 00212800013e9313 00212800013e9314 00212800013e93f7 FabMan gateway_name gt 3 Or remove ports from the partition FabMan gateway_name gt smpartition remove n partition_name pkey p_key port port ALL_CAS ALL SWITCHES ALL_ROUTERS See Step 2 for variable definitions For example FabMan gateway
180. ough the Gateway on page 7 Troubleshooting the Gateway 11 12 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Understanding Administrative Commands These topics provide an overview of administrative tasks and the command sets to perform those tasks Administering the gateway requires accessing the management controller Configuration Overview on page 13 m Oracle ILOM Command Overview on page 16 m Hardware Command Overview on page 16 a InfiniBand Command Overview on page 17 Related Information Troubleshooting the Gateway on page 1 Administering the Chassis on page 19 Administering the 14 Switch Chip on page 31 Administering Gateway Resources on page 137 Administering the InfiniBand Fabric on page 83 Administering the Subnet Manager on page 49 Configuration Overview After you have installed the gateway you can configure its functionality This table lists functions of the gateway to configure in the order of dependency If you do not want to configure a particular functionality skip that step Step Functionality Description Links 1 Configure the Configure the priority Set the Subnet Manager Priority on page 56 Subnet Manager Configure the prefix Set the Subnet Manager Prefix on page 57 Enable controlled handover Enable Subnet Manager Controlled Handover
181. ove the power cord and affected power supply from the gateway chassis Refer to Gateway Service powering off the power supply and Gateway Service removing the power supply b Wait one minute c Reinstall the power supply Refer to Gateway Service installing the power supply d Reattach the power cord Refer to Gateway Service powering on the power supply e If the power supply s Attention LED lights or the checkpower command still reports Alert or Not Present for the power supply replace the power supply Refer to Gateway Service servicing the power supplies Related Information m Gateway Reference checkpower command m Check Board Level Voltages on page 21 a Display Gateway Environmental and Operational Data on page 23 V Check Board Level Voltages When you want to know if the voltages used by the various components within the gateway are within nominal values use the checkvoltages command If a voltage deviates more than 10 of its nominal value there is a problem This check also reports if a battery has failed On the management controller type FabMan gateway_name gt checkvoltages Voltage ECB OK Measured 3 3V Main 3 30 V Measured 3 3V Standby 3 42 V Measured 12V 12 06 V Measured 5V 5 03 V Measured VBAT 3 17 V Measured 1 0V 1 01 V Measured I4 1 2V 1 22 V Measured 2 5V 2 51 V Measured V1P2 DIG 1 18 V Measured V1P2 ANG 1 18 V Measured 1 2V BridgeX 1 22 V
182. page 24 a Display Free Memory and File System Space on page 25 a Display Chassis FRU ID on page 26 a Display Power Supply FRU ID on page 26 a Display Gateway Firmware Versions on page 27 19 a Oracle Auto Service Request on page 28 Related Information m Monitoring the 14 Switch Chip on page 31 m Monitoring Gateway Resources on page 172 m Monitoring the InfiniBand Fabric on page 83 m Monitoring the Subnet Manager on page 49 m Controlling the Chassis on page 28 Vv Display Gateway General Health An easy way to perform a quick check of the gateway s operation is with the showunhealthy command On the management controller type FabMan gateway_name gt showunhealthy OK No unhealthy sensors FabMan gateway_name gt Related Information m Gateway Reference showunhealthy command a Display Gateway Environmental and Operational Data on page 23 V Display Power Supply Status The checkpower command performs a simple pass fail test on the power supplies 1 On the management controller type FabMan gateway_name gt checkpower PSU 0 present status OK PSU 1 present status Alert FabMan gateway_name gt 2 If you see the words Alert or Not Present in the command output perform these steps for the affected power supply 20 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 a Rem
183. page 33 Display Switch Chip Boot Status on page 34 Display Link Status on page 35 Display Connectors Configured With Autodisable Functionality on page 36 Display Switch Chip Port Status on page 37 31 a Display Switch Chip Port Counters on page 38 Related Information m Monitoring the Chassis on page 19 m Monitoring Gateway Resources on page 172 m Monitoring the InfiniBand Fabric on page 83 m Monitoring the Subnet Manager on page 49 m Controlling the I4 Switch Chip on page 40 V Display the Switch Chip Port to QSFP Connector Mapping You can use the dcsport command to display the mapping of a switch chip port to its respective QSFP connector and to display the mapping of a QSFP connector to its switch chip port On the management controller type FabMan gateway_name gt decsport port port where port is the number of the port 1 to 36 Alternatively type FabMan gateway_name gt dcsport connector connector where connector is the number of the QSFP connector 0A to 15B For example FabMan gateway_name gt dcsport port 7 DCS GW Switch port 7 maps to connector 14A FabMan gateway_name gt dcsport connector 14A Connector 14A maps to Switch port 7 FabMan gateway_name gt Note For a complete mapping of all ports to all connectors use the dcsport printconnnectors command 32 Sun Network QDR InfiniBand Gateway Switch Admi
184. parameters See Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 5 Create the appropriate configuration files See m Create the Central Configuration File Linux on page 154 m Create the VNIC Specific Configuration Files Linux on page 156 6 Log in as superuser of the host and start the mlx4_vnic_confd daemon etc init d mlx4_vnic_confd start The VNICs are created 7 Repeat Step 6 for all hosts of the InfiniBand fabric Related Information m Configure and Create VNICs for Gateway Manual Mode Linux on page 146 m Host Manual Mode Overview Linux on page 149 m Central Configuration File Linux on page 150 m VNIC Specific Configuration File Linux on page 151 Administering Gateway Resources 159 Determine VNIC Configuration Parameters for Host Manual Mode Linux on page 152 Create the Central Configuration File Linux on page 154 Create the VNIC Specific Configuration Files Linux on page 156 mlx4_vnic_confd Daemon Linux on page 157 Creating Virtual IO Adapters Oracle Solaris These topics discuss steps to enable gateway functionality for Oracle Solaris 11 hosts Step Description Links 1 Review background Ethernet Over InfiniBand Overview Oracle information Solaris on page 161 2 Verify the driver is installed Verify That the Ethernet Over InfiniBand Driver Is Installed Oracl
185. partition key Related Information m Gateway Manual Mode Overview Linux on page 143 m Determine VNIC Configuration Parameters For Gateway Manual Mode Linux HA on page 144 m Configure and Create VNICs for Gateway Manual Mode Linux on page 146 m VNIC Configuration on Linux Hosts in Gateway Manual Mode on page 147 V Configure and Create VNICs for Gateway Manual Mode Linux 1 Determine the VNIC configuration parameters See Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 2 Determine VLAN associations See Determine VLAN Associations for Gateway Manual Mode Linux on page 145 146 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 3 Create VLANs See Create VLANs on page 192 4 Create VNICs See Create VNICs on page 195 Related Information m Configure and Create VNICs for Host Manual Mode Linux on page 158 m Gateway Manual Mode Overview Linux on page 143 m Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 m Determine VLAN Associations for Gateway Manual Mode Linux on page 145 a VNIC Configuration on Linux Hosts in Gateway Manual Mode on page 147 VNIC Configuration on Linux Hosts in Gateway Manual Mode If you are creating VNICs in Gateway Manual Mode you must meet these conditions VNIC names appe
186. portive Software Linux on page 137 m Creating VNICs Under Gateway Manual Mode Linux on page 142 m Creating VNICs Under Host Manual Mode Linux on page 148 m Creating Virtual IO Adapters Oracle Solaris on page 160 m Monitoring Gateway Resources on page 172 m Controlling LAGs on page 183 m Controlling VLANs and VNICs on page 190 Controlling Gateway Ports and Parameters on page 202 Related Information m Troubleshooting the Gateway on page 1 m Understanding Administrative Commands on page 13 m Administering the Chassis on page 19 m Administering the 14 Switch Chip on page 31 m Administering the InfiniBand Fabric on page 83 m Administering the Subnet Manager on page 49 Installing Gateway Supportive Software Linux Install the BXOFED software on each host of the InfiniBand fabric that utilizes the VNIC and VLAN features of the gateway Within the BXOFED software package are drivers and applications you use to configure VNICs and VLANs on the hosts m Acquire the BXOFED Software Linux on page 138 137 a Install the BXOFED Software Linux on page 139 m Acquire the ConnectX 2 Firmware on page 141 Related Information m Creating VNICs Under Gateway Manual Mode Linux on page 142 m Creating VNICs Under Host Manual Mode Linux on page 148 m Creating Virtual IO Adapters Oracle Solaris on page 160 m
187. r Whether the Fabric Director instance on the fabric element is the master Related Information m Gateway Reference dconfig command m Display the InfiniBand Fabric Configuration on page 94 m Fabric Director and Fabric Elements on page 125 m Create a Fabric Configuration on page 127 m Add or Remove an Element From the Fabric Configuration on page 129 m Modify an Element of the Fabric Configuration on page 131 Create a Fabric Configuration Before you can use the fabric MIBs you must first create a fabric configuration You must create identical an configuration on all elements of the fabric 1 On the management controller initiate a fabric configuration session FabMan gateway_name gt f dconfig start fabric config empty FabMan gateway_name gt 2 Define the fabric name FabMan gateway_name gt fdconfig define fabric name name where name is the identifier of the fabric For example FabMan gateway_name gt fdconfig define fabric name test Fabric name updated FabMan gateway_name gt 3 Define an element of the fabric FabMan gateway_name gt fdconfig define element name name ip IP_address type type role rolel role role2 where Administering the InfiniBand Fabric 127 a name is the host name of the element as configured by Oracle ILOM m IP_address is the IP address of an element m type is the identifier of the type of element sw36 S
188. ration 1 Verify the smnodes list See Display the smnodes List on page 95 2 Initiate a fabric element configuration session on the management controller FabMan gateway_name gt createfabric start from current FabMan gateway_name gt 3 Display the fabric element configuration FabMan gateway_name gt createfabric list modified Hostname ILOM IP Addr Redundant IP Addr mnm12 gw 1 123 45 67 89 mnm13 gw 1 123 45 67 90 FabMan gateway_name gt Administering the InfiniBand Fabric 123 4 Modify a fabric element of the configuration FabMan gateway_name gt createfabric modify element name hostname ilomIp I P_address redundantIP P_address where m hostname is the host name of the element as seen in the Hostname column m IP_address is the IP address for the element recognized by Oracle ILOM or IPoIB For example FabMan gateway_name gt createfabric modify element name mnm13 gw 1 ilomIp 123 45 67 91 Element modified FabMan gateway_name gt 5 Verify the fabric element configuration FabMan gateway_name gt createfabric list modified Hostname ILOM IP Addr Redundant IP Addr mnm12 gw 1 123 45 67 89 mnm13 gw 1 123 45 67 91 FabMan gateway_name gt 6 End the configuration session and commit the in progress configuration to the active fabric element configuration FabMan gateway_name gt createfabric complete FabMan gateway_name gt 7 Repeat S
189. rch window reformats Click Read Me to display the README file Click Download The File Download window opens Click the filename zip link to initiate the download For example p12621910_151 Linux x86 64 zip Indicate where the file should be saved The file is downloaded and saved In your receiving directory decompress the filename zip file The BXOFED software is in the BXOFED 1 5 1 version_for Oracle tgz file There are also README release notes installation guide and user manual files in the filename zip file Read the README release notes and installation guide files for information on how to install the BXKOFED software Related Information Gateway Remote Management acquiring the gateway firmware package Install the BXOFED Software Linux on page 139 V Install the BXOFED Software Linux When you install the BXOFED software any previous installations of OFED or BXOFED software are removed Configuration files are not removed Administering Gateway Resources 139 Note If you are installing the BXOFED software on a cluster install the software onto one of the cluster nodes then install the rpm files in the OFED 1 5 1 RPMS on all remaining cluster nodes using cluster aware tools 1 Become superuser of the host that received the BXOFED software 2 Change to the directory where you extracted the tgz file 3 Run the installation script BXOFED 1 5 1 1 6 3 install pl
190. re than one MAC address can be assigned to a VIOA the VNIC management commands of the gateway are actually managing the MAC addresses within the gateway itself Related Information m Verify That the Ethernet Over InfiniBand Driver Is Installed Oracle Solaris on page 162 m Install the Ethernet Over InfiniBand Driver Oracle Solaris on page 162 m Set Up VIOAs Oracle Solaris on page 163 m Configure IP Addresses on the Data Link Oracle Solaris on page 167 m Create Oracle Solaris VNICs With VLAN Tagging Oracle Solaris on page 169 Administering Gateway Resources 161 V Verify That the Ethernet Over InfiniBand Driver Is Installed Oracle Solaris 1 Become superuser of the Oracle Solaris 11 host 2 Type pkg info ethernet over ib Name system io infiniband ethernet over ib Summary Ethernet over InfiniBand EoIB Drivers Description InfiniBand device driver implementing Ethernet over InfiniBand Category System Hardware State Installed Publisher solaris Version 0 5 11 Build Release 5 11 Branch 0 175 1 0 0 2 17991 Packaging Date Mon Oct 24 10 50 29 2011 Size 304 74 kB FMRI pkg solaris system io infiniband ethernet over ib 0 5 11 5 11 0 175 1 0 0 2 17991 20111024T105029Z 3 If there is no information returned or an error message is displayed install the Ethernet over InfiniBand driver See Install the Ethernet Over InfiniBand Driver Oracle Solaris on page
191. reate LAGs on page 185 m Add or Delete Connectors From a LAG on page 187 m Delete a LAG on page 188 a Disable LAG Mode on page 189 184 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 V Create LAGs Once LAG mode is enabled you can create LAGs 1 Determine which connectors will be grouped into a LAG and what the LAG names will be Note A connector is assigned to only one LAG at a time Note If a VNIC has been assigned to a connector that connector cannot be added to a LAG until the respective VNIC has been deleted 2 On the management controller create the first LAG FabMan gateway_name gt createlag lagname discoverypkey p_key datasl sl connector connector2 connectorN n lacpmode active passive off vniclimit 1 4 where m lagname is the identifier of the LAG connector is the name of the first connector to be added to the LAG 0A ETH 1 to OA ETH 4 and 1A ETH 1 to 1A ETH 4 m connectorN is the name of the last connector to be added to the LAG OA ETH 1 to OA ETH 4 and 1A ETH 1 to 1A ETH 4 m p_key is the discovery partition key 1 to 7 or default m si is the data service level Note Use service level 1 s1 1 only for low latency high priority small message low bandwidth traffic Use other service levels for regular high bandwidth traffic Note The
192. reated in firmware versions 2 0 or 2 1 are functional under either firmware versions 80 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Related Information m Add a New Subnet Manager to the Secret M_Key Fabric on page 78 m Merge Subnets Into One Secret M_Key Fabric on page 79 Administering the Subnet Manager 81 82 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Administering the InfiniBand Fabric These topics describe the administration of the InfiniBand fabric m Monitoring the InfiniBand Fabric on page 83 m Controlling the InfiniBand Fabric on page 97 Related Information m Troubleshooting the Gateway on page 1 m Understanding Administrative Commands on page 13 m Administering the Chassis on page 19 m Administering the 14 Switch Chip on page 31 m Administering the Subnet Manager on page 49 m Administering Gateway Resources on page 137 Monitoring the InfiniBand Fabric These topics enable you to display and check the operation and status of the InfiniBand fabric and components Note To use all the commands described in these topics you must access them through the SYS Fabric_Mgmt Linux shell target of the Oracle ILOM CLI m Displaying Information About Nodes on page 84 m Displaying Fabric Functionality Configuration Information
193. redefined partitions using the smpartition add command the added ports take precedence over the default ALL_SWITCHES and ALL_CAS port groupings The SELF and ALL port groupings are not supported by the smpartition command To specify all ports for a partition use the ALL_SWITCHES port grouping in conjunction with the ALL_CAS port grouping in the smpartition add command The ALL_SWITCHES port grouping must have full membership in the default partition P_Key 0x7fff The ALL_CAS port grouping can have limited membership in the default partition P_Key 0x7 To remove the ipoib flag from a partition use the smpartition modify flag command without any additional arguments If you use EoIB and VNIC functionality in a non default partition you must specify the appropriate BridgeX chips InfiniBand port GUIDs to the partition When you know the VNIC s Ethernet connector this table provides the respective BridgeX chip s portname Ethernet Connector BridgeX Portname OA ETH 1 Bridge 0 2 OA ETH 2 Bridge 0 2 OA ETH 3 Bridge 0 1 OA ETH 4 Bridge 0 1 1A ETH 1 Bridge 1 2 Administering the InfiniBand Fabric 107 Ethernet Connector BridgeX Portname 1A ETH 2 Bridge 1 2 1A ETH 3 Bridge 1 1 1A ETH 4 Bridge 1 1 When you know the Portname you can look in the output of the showgwports command for the respective PortGUID Using the table and the showgwports command together when you know the Ethernet con
194. rence enablevnic command m Gateway Reference showvnics command a VNIC Allocation on the Gateway on page 191 Administering Gateway Resources 199 m Create VLANs on page 192 m Delete VLANs on page 194 m Create VNICs on page 195 a Disable VNICs on page 200 m Delete VNICs on page 201 m Display the VNICs on page 176 Y Disable VNICs If you believe that there is a problem with a VNIC you can disable it 1 Identify the VNIC you want to disable See Display the VNICs on page 176 2 Disable the VNIC on the management controller FabMan gateway_name gt disablevnic connector vnic_ID where m connector is the name of the connector OA ETH 1 to 0A ETH 4 0A ETH 1A ETH 1 to 1A ETH 4 and 1A ETH m vnic_ID is the VNIC identifier For example to disable VNIC 15 on connector 1A ETH 1 type FabMan gateway_name gt disablevnic 1A ETH 1 15 FabMan gateway_name gt 3 Repeat from Step 1 for all VNICs to be disabled 4 Verify that the VNICs have been disabled FabMan gateway_name gt showvnics ID STATE FLG IOA_GUID NODE IID MAC VLN PKEY GW 16 UP N 00 21 28 00 01 3E CE A0 mnm34 55 0000 02 02 02 02 02 04 3 ffff 1A ETH 4 13 UP N 00 03 BA 00 01 00 E3 71 mnm34 60 0000 02 02 02 02 02 01 3 ffff 1A ETH 3 200 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 15 DISABLED N 00 21 28 0
195. ring Gateway Resources 141 11 12 13 14 15 16 17 18 Click Search The Patch Search window expands with the search results In the Patch Name column click the respective patch number link For example 12610332 The Patch Search window reformats Click Read Me to display the README file Click Download The File Download window opens Click the filename zip link to initiate the download For example p12610332__Solaris86 64 zip Indicate where the file should be saved The file is downloaded and saved In your receiving directory decompress the filename zip file The ConnectX 2 firmware is in the fw ConnectX2 rel 2_7_version bin file For example fw ConnectX2 rel 2_7_8130 375 3697 01 bin Refer to your HCA documentation for instructions on how to upgrade the ConnectX 2 firmware Creating VNICs Under Gateway Manual Mode Linux These topics describe how to create VNICs under gateway manual mode Gateway Manual Mode Overview Linux on page 143 Determine VNIC Configuration Parameters For Gateway Manual Mode Linux on page 144 Determine VLAN Associations for Gateway Manual Mode Linux on page 145 Configure and Create VNICs for Gateway Manual Mode Linux on page 146 VNIC Configuration on Linux Hosts in Gateway Manual Mode on page 147 Related Information Installing Gateway Supportive Software Linux on page 137 Creating VNICs Under Hos
196. ring Secret M_Keys on page 67 m Set the Replication Password on page 76 m Enable or Disable Secret M_Key Functionality on page 77 m Considering Situations In Regard to Secret M_Key on page 78 Vv Enable or Disable Secret M_Key Functionality Secret M_Key functionality is enabled or disabled independent of its configuration On the management controller that hosts the master Subnet Manager type FabMan gateway_name gt smsubnetprotection state where state is either enablesecretmkey or disablesecretmkey For example FabMan gateway_name gt smsubnetprotection enablesecretmkey FabMan gateway_name gt Related Information m Gateway Reference smsubnetprotection command m Secret M_Key Overview on page 64 m smsubnetprotection Command on page 65 m Prepare for Secret M_Key Functionality on page 66 Administering the Subnet Manager 77 Configuring Secret M_Keys on page 67 m Set the Replication Password on page 76 m Enable or Disable Secret M_Key Persistence on page 76 m Considering Situations In Regard to Secret M_Key on page 78 Considering Situations In Regard to Secret M_Key When secret M_Key functionality is enabled certain situations require additional considerations m Add a New Subnet Manager to the Secret M_Key Fabric on page 78 m Merge Subnets Into One Secret M_Key Fabric on page 79 m Secret M_Key and Firmware Upgrade and Down
197. rity is 0 lowest to 13 highest For example FabMan gateway_name gt setsmpriority 5 Current SM settings smpriority 5 controlled_handover FALSE subnet_prefix 0xfe80000000000000 M_Key None Routing engine FatTree FabMan gateway_name gt 3 Enable the Subnet Manager See Enable the Subnet Manager on page 62 Related Information Gateway Reference setsmpriority command Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 Set the Subnet Manager Prefix on page 57 Enable Subnet Manager Controlled Handover on page 58 Set the Subnet Manager Readable Management Key on page 59 Clear the Subnet Manager Readable Management Key on page 60 Set the Subnet Manager Routing Algorithm on page 61 Enable the Subnet Manager on page 62 Disable the Subnet Manager on page 63 Set the Subnet Manager Prefix The setsubnetprefix command writes a prefix value to the subnet_prefix parameter of the configuration file 1 On the management controller disable the Subnet Manager See Disable the Subnet Manager on page 63 Administering the Subnet Manager 57 2 Set the Subnet Manager prefix FabMan gateway_name gt setsubnetprefix Oxabbababe Current SM settings smpriority 5 controlled_handover FALSE subnet_prefix Oxabbababe M_Key None Routing engine FatTree FabMan gateway_name gt
198. rovide access to or information on content products and services from third parties Oracle Corporation and its affiliates are not responsible for and expressly disclaim all warranties of any kind with respect to third party content products and services Oracle Corporation and its affiliates will not be responsible for any loss costs or damages incurred due to your access to or use of third party content products or services Copyright 2013 Oracle et ou ses affili s Tous droits r serv s Ce logiciel et la documentation qui Laccompagne sont pee par les lois sur la propri t intellectuelle Ils sont conc d s sous licence et soumis des restrictions d utilisation et de divulgation Sauf disposition de votre contrat de licence ou de la loi vous ne pouvez pas copier reproduire traduire diffuser modifier breveter transmettre distribuer exposer ex cuter publier ou afficher le logiciel m me partiellement sous quelque forme et par uelque proc d que ce soit Par ailleurs il est interdit de proc der a toute ing nierie inverse du logiciel de le d sassembler ou de le d compiler except a es fins d interoperabilit avec des logiciels tiers ou tel que prescrit par la loi Les informations fournies dans ce document sont susceptibles de modification sans pr avis Par ailleurs Oracle Corporation ne garantit pas qu elles soient exemptes d erreurs et vous invite le cas ch ant a lui en faire part par crit
199. rting FAN test Fan 0 not present Fan 1 running at rpm 12075 Fan 2 running at rpm 11960 Fan 3 running at rpm 12075 Fan 4 not present FAN test returned OK Starting Connector test Connector test returned OK Starting Onboard ibdevice test Switch OK Bridge 0 OK Bridge 1 OK All Internal ibdevices OK Onboard ibdevice test returned OK Starting SSD test SSD test returned OK Environment test PASSED FabMan gateway_name gt Related Information E Gateway Reference env_test command Vv Display Flash Drive Information Included in the output of the showdisk command is the estimated remaining life of the SSD drive which can help in troubleshooting On the management controller type FabMan gateway_name gt showdisk Device Model 512MB NANDrive Serial Number 0000000000A1PsB87FAa Firmware Version ND B642J 24 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 SMART supported yes SMART enabled yes Average age 4 groups group 0 12817 group 1 533 group 2 TL5 group 3 954 Block erase count 4 groups group 0 12782181 group 1 532378 group 2 FL3173 group 3 952668 Total 14980400 Total physical blocks 4096 Estimated Life Usage 12 817 Estimated Remaining Life 87 183 FabMan gateway_name gt Related Information m Gateway Reference showdisk command a Display Gateway Environmental and Operational Dat
200. s for Host Manual Mode Linux on page 158 9 Save the Back up the configuration for Gateway Remote Management backing up the configuration restoration at a later time configuration should it become necessary Related Information m Oracle ILOM Command Overview on page 16 m Hardware Command Overview on page 16 a InfiniBand Command Overview on page 17 Understanding Administrative Commands 15 Oracle ILOM Command Overview The Oracle ILOM CLI web SNMP and IPMI interfaces enable additional administration features and capabilities The Oracle ILOM CLI SNMP and IPMI interfaces use remote clients for command line administration of many Oracle ILOM targets Alternatively the Oracle ILOM web interface permits point and click administration of the Oracle ILOM components and services After you log in to the ilom admin account the Oracle ILOM prompt gt appears and you can enter Oracle ILOM commands in this format gt command option target property value Information about Oracle ILOM support of the gateway is available in the Oracle Integrated Lights Out Manager ILOM 3 0 Supplement for the Sun Network QDR InfiniBand Gateway Switch available online at http www oracle com pls topic lookup ctx E19671 01 amp id homepage Overall information about Oracle ILOM 3 0 is available online at http www oracle com pls topic lookup ctx E19860 01 amp id homepage Related Information
201. s is the IP address of the interface For example FabMan gateway_name gt create_ipoib n newipoib m 255 255 254 0 t both 0x0001 123 45 67 89 Create interface Name PKey IP address Subnetmask Interface Interface type newipoib 0x0001 123 45 67 89 255 255 254 0 ibd both FabMan gateway_name gt Related Information m Gateway Reference create_ipoib command m Display the IPoIB Interface on page 96 m Delete the IPoIB Interface on page 136 m Prepare for Secret M_Key Functionality on page 66 V Delete the IPoIB Interface On the management controller type FabMan gateway_name gt delete_ipoib if_name p_key where if_name is the identifier of the IP interface m p_key is the partition key 1 to 7 f or default For example FabMan gateway_name gt delete_ipoib newipoib Delete interface Name PKey IP address Subnetmask Interface Interface type newipoib 0x0001 123 45 67 89 255 255 254 0 ibo both FabMan gateway_name gt Related Information m Gateway Reference delete_ipoib command m Display the IPoIB Interface on page 96 m Create the IPoIB Interface on page 135 136 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Administering Gateway Resources These topics describe the host and BridgeX functionality of the gateway and how to configure monitor and control that functionality m Installing Gateway Sup
202. s the management key 16 hexadecimal digits optional m lid is the LID of the node m port is the port of the node For example FabMan gateway_name gt ibportstate 15 23 reset Initial PortInfo Port info Lid 15 port 23 LankState ican s t cece es oe eet ae DOWN PhysLinkState Disabled LinkWidthSupported 1X or 4X 102 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 LinkWidthEnabled 1X or 4X LinkWidthActive 2 0000 4X LinkSpeedSupported 4 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedEnabled 4 2 5 Gbps or 5 0 Gbps or 10 0 Gbps LinkSpeedActive 000 eee 2 5 Gbps After PortInfo set Port info Lid 15 port 23 LINnKkState tcc ated ee as Down PhyshankState s ccdd bedeneeiane Disabled After PortInfo set Port info Lid 15 port 23 LInkStates eraa e aaa eee ee bee cla wees Down PhyslinkState lt 4 a6 cig Ei Seed eae wd PortConfigurationTraining FabMan gateway_name gt Related Information m Gateway Reference ipportstate command m Disable or Enable an External Port on page 104 Set Port Speed You can manually set the speed of a single port to help determine symbol error generation The ibportstate command can set the speed to 2 5 5 0 or 10 0 Gbit sec Note If secret M_Key functionality is enabled use the M m_key option to t
203. s usually reserved for a switch or bridge management VLAN m name is the name of the VNIC For ease of identification the VNIC name can be formatted as eX _vnicY where a X is the instance number of the Ethernet over InfiniBand IP interface a Y is the instance of the VNIC independent of the interface instance For example dladm create vnic 1l net6 m 2 8 20 7c 5d d3 v 3 e0_vnic0 5 Verify the creation of the VNIC dladm show vnic LINK OVER SPEED MACADDRESS MACADDRTYPE VID e0_vnicO net6 10000 2 8 20 7c 5d d3 random 3 6 Display a list of all data links dladm show link LINK net6 e0_vnicO 7 Configure the IP address of the data link e0_vnicO See Configure IP Addresses on the Data Link Oracle Solaris on page 167 Administering Gateway Resources 171 8 On the management controller verify that the VNICs are properly associated and up FabMan gateway_name gt showvnics ID STATE FLG IOA_ GUID NODE IID MAC VLN PKEY GW 0 UP N 0021280001CF37AC solaris01 00c0 02 42 42 42 12 03 NO Oxffff 0A ETH 1 1 UP H 0021280001CF37AC solaris01 80c1 02 08 20 7C 5D D3 3 Oxffff OA ETH 1 FabMan gateway_name gt Related Information dladm command man page Ethernet Over InfiniBand Overview Oracle Solaris on page 161 Verify That the Ethernet Over InfiniBand Driver Is Installed Oracle Solaris on page 162 Install the Et
204. smnodes List 95 vi Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 v v v Display the Fabric Element Configuration 96 Display the IPoIB Interface 96 Display Configured Secret M_Keys 97 Controlling the InfiniBand Fabric 97 Investigating Nodes 98 v v v v Perform Comprehensive Diagnostics for the Entire Fabric 98 Determine Changes to the InfiniBand Fabric Topology 99 Find 1x SDR or DDR Links in the Fabric 100 Determine Which Links Are Experiencing Significant Errors 100 Controlling Nodes 101 v v v v Clear Data and Error Counters 102 Reset a Port 102 Set Port Speed 103 Disable or Enable an External Port 104 Partitioning the InfiniBand Fabric 105 smpartition Command 106 smpartition Command Guidelines 107 Partitions and P_Keys 108 lt a lt a a v Determine the Partitions and P_Keys 109 Create a User Partition 110 Add or Remove a Port From a Partition 112 Modify a Partition or Port 114 Delete a Partition 116 Remove User Partitions for Firmware Downgrade 117 Correlating Fabric Elements 119 createfabric Command 120 v v Create a Fabric Element Configuration 120 Add or Delete an Element From the Configuration 122 Contents vii v Modify an Element of the Configuration 123 Configuring the Fabric Director Node List 125 Fabric Director and Fabric Elements 125 dconfig Command Overview 126 v Create a Fabric Confi
205. splay the InfiniBand Fabric Topology Detailed on page 86 Vv Display the InfiniBand Fabric Configuration If you have configured the InfiniBand fabric for SNMP operations you can display the elements configured with the fdconfig command 94 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 1 On the management controller type FabMan gateway_name gt f dconfig list current fabric config Name IP Addr Type Role s isMaster Fabricname primary 123 45 67 89 sw36gw gateway test secondary 123 45 67 90 sw36 leafSwitch no test FabMan gateway_name gt In this example the element named primary has IP address 123 45 67 89 and is a Sun Network QDR InfiniBand Gateway Switch It is being used as a gateway and is hosting the master Fabric Director daemon The other element named secondary has an IP address of 123 45 67 90 and is a Sun Datacenter InfiniBand Switch 36 It is being used as a leaf switch 2 If the information displayed is incorrect modify the fabric configuration See m Add or Remove an Element From the Fabric Configuration on page 129 m Modify an Element of the Fabric Configuration on page 131 Related Information m Gateway Reference dconfig command m Configuring the Fabric Director Node List on page 125 Vv Display the smnodes List 1 On the management controller type FabMan gateway_name gt smnodes list 123 45 67 89 123 45 67 9
206. sses 2 Create a numbered list one number per row 3 For each number provide a MAC address to the right Note Each MAC address must be unique and not 00 00 00 00 00 00 152 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Note Only even numbers are supported for the most significant byte of the MAC address unicast 4 For each MAC address open a terminal window and log in to the host to receive that MAC address 5 Use the ibv_devinfo command to determine the hca_id value and HCA port GUID For example ibv_devinfo hea_id m1x4_0 fw_ver 2 5 9266 node_guid 0003 ba00 0100 c708 sys_image_guid 0003 ba00 0100 c70b In the example the hca_id value is m1x4_0 and the HCA port GUID sys_image_guid is 003ba000100c70b 6 For each MAC address append the host name the respective hca_id value the port 1 or 2 and the HCA port GUID to the right of the MAC address 7 For each MAC address that will be associated with a VLAN including VLAN ID No append the VLAN identifier NO or 2 to 4094 to the right of the GUID Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN Note Due to hardware limitations for MultiCast groups there is a maximum of 1000 VLANs 8 For each MAC addr
207. t a Disable the port See Disable a Switch Chip Port on page 40 b Reenable the port Do not use the automatic option when you reenable the port See Enable a Switch Chip Port on page 41 Related Information m Gateway Reference enableswitchport command m Autodisable Functionality on page 43 a Add Connectors to the Autodisable List on page 44 m Delete Connectors From the Autodisable List on page 46 V Delete Connectors From the Autodisable List On the management controller type FabMan gateway_name gt autodisable del connector cause where m connector is the name of the connector 0A to 15A and OB to 15B m cause is the reason to disable the connector a H The error rate is higher than what the InfiniBand specification permits 46 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 a L The link speed or width is less than optimal less than 4x QDR Alternatively to delete all connectors type FabMan gateway_name gt autodisable none In this situation all connectors are deleted from the autodisable list regardless of cause For example to delete connector 0A from the autodisable list for suboptimal link width FabMan gateway_name gt autodisable del 0A L FabMan gateway_name gt Related Information m Gateway Reference autodisable command m Autodisable Functionality on page 43 m Add Con
208. t Manual Mode Linux on page 148 142 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m Creating Virtual IO Adapters Oracle Solaris on page 160 m Monitoring Gateway Resources on page 172 m Controlling LAGs on page 183 m Controlling VLANs and VNICs on page 190 Controlling Gateway Ports and Parameters on page 202 Gateway Manual Mode Overview Linux The default and suggested means of creating and managing VNICs on the gateway is through use of BridgeX gateway administrative commands To use the commands described in these topics you must access them through the SYS Gateway_Mgmt or SYS Fabric_Mgmt Linux shell targets of the Oracle ILOM CLI The VNICs you create through the commands are based on the InfiniBand hosts When creating VNICs consider this m You must specify the MAC for the VNIC If you do not the default MAC for the VNIC is 00 00 00 00 00 00 which renders the VNIC unusable m You must ensure that each MAC is unique a If you do not have a list of unique global MACs use locally administrated MACs a VNICs will not go to an up state without an active Subnet Manager present m Manually created VNICs are persistent and survive reboots and power cycles When creating VNICs consider that there are conditions from partitioning See smpartition Command Guidelines on page 107 Related Information m Host Manual Mode Overview
209. tain the restricted Linux shell prompt FabMan gateway_name gt You can now enter InfiniBand commands in this format FabMan gateway_name gt command option option Related Information m Gateway Reference understanding InfiniBand commands m Oracle LOM Command Overview on page 16 a Hardware Command Overview on page 16 Understanding Administrative Commands 17 18 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Administering the Chassis These topics describe the administration of the gateway m Monitoring the Chassis on page 19 m Controlling the Chassis on page 28 Related Information m Troubleshooting the Gateway on page 1 m Understanding Administrative Commands on page 13 m Administering the 14 Switch Chip on page 31 m Administering Gateway Resources on page 137 m Administering the InfiniBand Fabric on page 83 m Administering the Subnet Manager on page 49 Monitoring the Chassis These topics enable you to display and check the operation and status of the gateway m Display Gateway General Health on page 20 m Display Power Supply Status on page 20 m Check Board Level Voltages on page 21 m Display Internal Temperatures on page 22 m Display Fan Status on page 22 a Display Gateway Environmental and Operational Data on page 23 m Display Flash Drive Information on
210. tate 2000 LinkUp LinkDownDefState 0 Polling PHOLECEBIUS eid bs bebe hha es 0 DMG cague E ee A ee eee 0 Subnet Timeout s iad eee ei aaoi a ap da 0 RespTimeVa Let nc see e a a Soe ee ek 0 Hocal PRY SEL cs dese cas eos ed a ecece ace wl 8 OVE TUNED xc wos eE Er E eee kee E 8 MaxCredt Chant i badabe aa datas nace acs 85 ROUNAT IDSs rekta eke kee bea seas 16777215 FabMan gateway_name gt a Display Low Level Detailed Information About a Node on page 90 92 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Displaying Fabric Functionality Configuration Information These tasks display the configurations used for added functionality of the InfiniBand fabric m Display the InfiniBand Fabric Partition Configuration on page 93 m Display the InfiniBand Fabric Configuration on page 94 m Display the smnodes List on page 95 m Display the Fabric Element Configuration on page 96 m Display the IPoIB Interface on page 96 m Display Configured Secret M_Keys on page 97 Related Information a Displaying Information About Nodes on page 84 m Partitioning the InfiniBand Fabric on page 105 m Correlating Fabric Elements on page 119 m Configuring the Fabric Director Node List on page 125 m Performing Supportive Tasks on page 132 Vv Display the InfiniBand Fabric Partition Configuratio
211. te the IPoIB Interface on page 135 96 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 V Display Configured Secret M_Keys 1 On the management controller type FabMan gateway_name gt smsubnetprotection list active File format_version_number 1 Sun DCS IB mkey config file This file is generated do not edit secretmkey enabled nodeid o04nm2 gw 6 time 15 Sep 03 54 46 checksum 378d9b09744e1d8b8ba6ae868c99d0c9 commit_number 3 Mkey Untrusted Mkey Smkey Attribute 0x00abcdefabcdef01 Oxlaa45124fee612ae Ox15fc26aea300f831 0x00abcdefabcdef02 0x4ccd8230de6cd348 Ox3fc7e6ad701la8a2a 0x00abcdefabcdef03 0x9baaldebcc74de5e 0x1b253003600d137b C FabMan gateway_name gt 2 If the information displayed is incorrect reconfigure the secret M_Keys See Configuring Secret M_Keys on page 67 Related Information m Gateway Reference smsubnetprotection command m Set the Secret M_Key to Use on page 72 m Add Secret M_Keys to the Configuration on page 69 m Delete Secret M_Keys From the Configuration on page 73 Controlling the InfiniBand Fabric You can perform these tasks to manage the InfiniBand fabric and its components Note To use all the commands described in these topics you must access them through the SYS Fabric_Mgmt Linux shell target of the Oracle ILOM CLI m Investigating Nodes on page 98 m Controlling Nodes on page 101
212. tep 2 through Step 6 on all management controllers hosting Subnet Managers in the InfiniBand fabric Related Information m Gateway Reference createfabric command a Display the Fabric Element Configuration on page 96 m createfabric Command on page 120 m Create a Fabric Element Configuration on page 120 m Add or Delete an Element From the Configuration on page 122 124 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Configuring the Fabric Director Node List These topics enable you to configure the Fabric Director node list m Fabric Director and Fabric Elements on page 125 m fdconfig Command Overview on page 126 m Create a Fabric Configuration on page 127 m Add or Remove an Element From the Fabric Configuration on page 129 m Modify an Element of the Fabric Configuration on page 131 Related Information m Investigating Nodes on page 98 m Controlling Nodes on page 101 m Partitioning the InfiniBand Fabric on page 105 m Correlating Fabric Elements on page 119 m Performing Supportive Tasks on page 132 Fabric Director and Fabric Elements The Fabric Director monitors the InfiniBand fabric by polling the nodes or elements of the fabric Each fabric element has a Fabric Director instance and the two communicate with each other in a one to one relationship The Fabric Director gathers information
213. ters on page 205 m Set the Gateway Service Level on page 206 m Set the Gateway Instance Number on page 207 m Set the Gateway System Name on page 208 Vv Disable a Gateway Ethernet Port You can disable an enabled gateway Ethernet port with the disablegwport command On the management controller type FabMan gateway_name gt disablegwport connector where connector is the name of the connector hosting the port 0A ETH 1 to OA ETH 4 OA ETH 1A ETH 1 to 1A ETH 4 and 1A ETH For example to disable the port associated with connector 0A ETH 3 type FabMan gateway_name gt disablegwport 0A ETH 3 FabMan gateway_name gt Related Information m Gateway Reference disablegwport command m Enable a Gateway Ethernet Port on page 203 m Set the Gateway Ethernet Port Parameters on page 205 m Set the Gateway Service Level on page 206 m Set the Gateway Instance Number on page 207 204 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 m Set the Gateway System Name on page 208 V Set the Gateway Ethernet Port Parameters Use the setgwethport command to configure the gateway Ethernet ports On the management controller type FabMan gateway_name gt setgwethport connector mtu mtu txpause pause rxpause pause where m connector is the name of the connector OA ETH or 1A ETH m mtu is the value of the
214. teway_name gt getmaster 1 Local SM enabled and running state MASTER SM monitoring history 20130118 07 25 37 No Master SubnetManager seen in the system 20130118 07 25 55 Master SubnetManager on sm lid 1 sm guid 0x3baabbal15c0a0 SUN IB QDR GW switch o4nm2 gw 3 10 172 144 71 20130118 07 27 13 Master SubnetManager on sm lid 10 sm guid 0x21284694e5a0a0 SUN DCS 36P QDR o4nm2 36p 1 10 172 144 65 FabMan gateway_name gt Related Information Gateway Reference getmaster command Display Subnet Manager Status on page 50 Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 Verify the Subnet Manager Configuration Integrity on page 54 Display the Subnet Manager Log on page 55 Vv Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm If you want to know your Subnet Manager s priority controlled handover state prefix management key and routing algorithm use the list option of the setsmpriority command 1 On the management controller type FabMan gateway_name gt setsmpriority list Current SM settings smpriority 8 controlled_handover TRUE subnet_prefix 0xfe80000000000000 Administering the Subnet Manager 51 52 M_Key Oxabbababe0001 Routing engine FatTree FabMan gateway_name gt Note The management key displayed in the output is the readable M_Key and not
215. thX where X is identifier the name should be different from eth0 to eth7 since these names are reserved for devices created and controlled from the host A suggested name scheme would be ethgateway_instance_VNIC_ID but all names apart for eth0 to eth7 would be valid 5 Repeat Step 4 for all VNICs to be created 196 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 6 Verify the status and configuration of the VNICs FabMan gateway_name gt showvnics ID STATE FLG IOA_GUID NODE 16 UP ffff 1A 1 13 UP ffff 1A 1 15 UP ffff 1A 1 14 UP ffff 1A 1 ETH 2 N 00 21 28 00 01 3E CE A0O mnm34 55 N 00 03 BA 00 01 00 E3 71 mnm34 60 N 00 21 28 00 01 3E CE 9F mnm34 55 N 00 03 BA 00 01 00 E3 72 mnm34 60 FabMan gateway_name gt 0000 02 0000 02 0000 02 0000 02 02 02 02 02 02 02 02 02 02 02 02 02 02 02 Related Information Gateway Reference createvnic command Gateway Reference showvnics command VNIC Allocation on the Gateway on page 191 Create VLANs on page 192 Delete VLANs on page 194 Enable VNICs on page 199 Disable VNICs on page 200 Delete VNICs on page 201 Display the VNICs on page 176 Modify VNICs You can use the modifyvnic command to change the VNIC s MAC address or the VLAN with which the VNIC is associated
216. tionpassword password where password is eight alphanumeric characters For example FabMan gateway_name gt smsubnetprotection setreplicationpassword reindeer FabMan gateway_name gt Related Information Gateway Reference smsubnetprotection command Secret M_Key Overview on page 64 smsubnetprotection Command on page 65 Prepare for Secret M_Key Functionality on page 66 Configuring Secret M_Keys on page 67 Enable or Disable Secret M_Key Persistence on page 76 Enable or Disable Secret M_Key Functionality on page 77 Considering Situations In Regard to Secret M_Key on page 78 Vv Enable or Disable Secret M_Key Persistence When secret M_Key persistence is enabled the InfiniBand nodes retain the active secret M_Key should a power failure reset or reboot occur 76 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 On the management controller type FabMan gateway_name gt localmkeypersistence state where state is either enable or disable For example FabMan gateway_name gt localmkeypersistence enable FabMan gateway_name gt Related Information m Gateway Reference localmkeypersistence command m Verify M_Key Persistence on page 53 m Secret M_Key Overview on page 64 m smsubnetprotection Command on page 65 m Prepare for Secret M_Key Functionality on page 66 m Configu
217. tstatus command m Set the Gateway Ethernet Port Parameters on page 205 m Display LAG Information on page 173 178 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 a Display the VLANs on page 174 m Display the VNICs on page 176 a Display IO Adapters on page 177 a Display Gateway Port Information on page 179 m Display the Gateway Instance Number on page 181 a Display Information About the Gateway on page 182 Vv Display Gateway Port Information Use the showgwports command to display detailed information about the gateway ports Administering Gateway Resources 179 On the management controller type FabMan gateway_name gt showgwports v BRIDGE DEVICES 0x00212856d102c000 SUN IB QDR GW switch o4nm2 gw 6 10 172 144 77 Bridge 0 0x00212856d102c040 SUN IB QDR GW switch o4nm2 gw 6 10 172 144 77 Bridge 1 INTERNAL PORTS Device Port Portname PeerPort PortGUID LID IBState GWState Speed VLs MTU Bridge 0 1 Bridge 0 1 4 0x00212856d102c001 0x0002 Active Up 40Gbs 2 4096 Bridge 0 2 Bridge 0 2 3 0x00212856d102c002 0x0003 Active Up 40Gbs 2 4096 Bridge 1 1 Bridge 1 1 2 0x00212856d102c041 0x0004 Active Up 40Gbs 2 4096 Bridge 1 2 Bridge 1 2 1 0x00212856d102c042 0x0005 Active Up 40Gbs 2 4096 CONNECTOR OA ETH Port Bridge Adminstate Link State MTU TxPause RxPause Linkmode Speed OA ETH 1 Bridge 0 2 Enabled Up Up
218. ubnetprotection Command 65 v Prepare for Secret M_Key Functionality 66 Configuring Secret M_Keys 67 v Create the Secret M_Key Configuration 67 Contents v lt a lt v Add Secret M_Keys to the Configuration 69 Set the Secret M_Key to Use 72 Delete Secret M_Keys From the Configuration 73 Configure a Local Secret M_Key 75 v Set the Replication Password 76 v Enable or Disable Secret M_Key Persistence 76 v Enable or Disable Secret M_Key Functionality 77 Considering Situations In Regard to Secret M_Key 78 v v Add a New Subnet Manager to the Secret M_Key Fabric 78 Merge Subnets Into One Secret M_Key Fabric 79 Secret M_Key and Firmware Upgrade and Downgrade Guidelines 80 Administering the InfiniBand Fabric 83 Monitoring the InfiniBand Fabric 83 Displaying Information About Nodes 84 v DENE SEEE SE SE EEE EE v Identify All Switches in the Fabric 84 Identify All CAs in the Fabric 85 Display the InfiniBand Fabric Topology Simple 86 Display the InfiniBand Fabric Topology Detailed 86 Display a Route Through the Fabric 87 Display the Link Status of a Node 88 Display Counters for a Node 89 Display Low Level Detailed Information About a Node 90 Display Low Level Detailed Information About a Port 91 Displaying Fabric Functionality Configuration Information 93 v v v Display the InfiniBand Fabric Partition Configuration 93 Display the InfiniBand Fabric Configuration 94 Display the
219. ubnetprotection add 0x00abcdefabcdef03 Mkey added FabMan gateway_name gt 4 Repeat Step 3 for any additional M_Keys you want to add to the configuration 5 Optional If you will use one of the just added secret M_Keys set the M_Key to use FabMan gateway_name gt smsubnetprotection set current m_key where m_key is the management key 16 hexadecimal digits For example FabMan gateway_name gt smsubnetprotection set current 0x00abcdefabcdef03 Mkey 0x00abcdefabcdef03 updated FabMan gateway_name gt Note You can only set an M_Key which has been added to the configuration 70 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 6 Display the pending M_Keys FabMan gateway_name gt smsubnetprotection list modified File format_version_number 1 Sun DCS IB mkey config file This file is generated do not edit secretmkey enabled nodeid o04nm2 gw 6 time 15 Sep 03 47 12 checksum 0 commit_number 3 Mkey Untrusted Mkey Smkey Attribute 0x00abcdefabcdef01 Oxlaa45124fee612ae 0x15fc26aea300f831 0x00abcdefabcdef02 0x4ccd8230de6cd348 Ox3fc7e6ad701la8a2a C 0x00abcdefabcdef03 0x9baaldebcc74de5e 0x1b253003600d137b S FabMan gateway_name gt In the output the Attribute heading identifies m C The current secret M_Key m S The standby secret M_Key 7 End the session and commit the secret M_Key configuration to the master Subnet Manager
220. un Datacenter InfiniBand Switch 36 sw36gw Sun Network QDR InfiniBand Gateway Switch a unknown The element type is unknown m role is the purpose of the element gateway The element acts as a gateway spineSwitch The element acts as a spine switch m leafSwitch The element acts as a leaf switch unknown The element role is unknown For example FabMan gateway_name gt fdconfig define element name primary ip 123 45 67 89 type sw36gw role leafSwitch role gateway Element added FabMan gateway_name gt 4 Verify the fabric configuration FabMan gateway_name gt fdconfig list in progress fabric config Name IP Addr Type Role s isMaster Fabricname primary 123 45 67 89 sw36gw gateway test FabMan gateway_name gt 5 Consider your next step If you want to add elements perform Step 3 to Step 4 for each new element a If you want to abort the configuration altogether type FabMan gateway_name gt dconfig abort In progress config aborted FabMan gateway_name gt m Otherwise go to Step 6 6 End the configuration session and commit to the new configuration FabMan gateway_name gt f dconfig complete fabric config FabMan gateway_name gt 128 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 7 Perform Step 1 through Step 6 on the management controllers of all elements of the fabric The configuratio
221. unable to access the management controller power cycle the gateway Refer to Gateway Service powering off the power supply and Gateway Service powering on the power supply If you are able to access the management controller restart the management controller See Restart the Management Controller on page 29 Check overall gateway health See Display Gateway General Health on page 20 Verify that the gateway is within operating temperatures and voltages See Display Gateway Environmental and Operational Data on page 23 If the previous steps do not rectify the situation replace the gateway Refer to Gateway Service removing the gateway from the rack Refer to Gateway Installation installing the gateway into the rack Check the fan speed See Display Fan Status on page 22 If the previous step does not rectify the situation replace the fan Refer to Gateway Service servicing fans If the previous step does not rectify the situation install the fan into another available slot Refer to Gateway Service servicing fans If no other slots are available replace the gateway Refer to Gateway Service removing the gateway from the rack Refer to Gateway Installation installing the gateway into the rack 2 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 March 2013 Related Information InfiniBand Fabric Problems on page 3 Net
222. vnic command 201 deleting autodisabled connectors 46 connectors from LAGs 187 fabric elements from configuration 122 IPoIB interface 136 LAGs 188 partition configuration 116 secret M_Keys 73 smnodes from list 134 VLANs 194 VNICs 201 dellagport command 187 disablegwport command 204 disablelagmode command 189 disablesmcommand 63 disableswitchport command 40 disablevnic command 200 disabling gateway Ethernet port 204 LAG mode 189 port 104 secret M_Key functionality 77 persistence 76 Subnet Manager 63 switch chip port 40 VNICs 200 displaying autodisable configured connectors 36 Index 211 212 disabled links 35 fabric configuration 94 fabric configuration information 93 fabric elements 96 file system space 25 flash drive information 24 free memory 25 FRU ID chassis 26 power supply 26 gateway Ethernet information 178 firmware 27 health 20 information 182 IO adapters 177 port information 179 service level 179 182 system name 182 instance number 181 IPoIB interface 96 LAGs 173 M_Key persistence 53 node counters 89 information 84 90 link status 88 partition configuration 93 port information 91 route 87 secret M_Key configuration 52 persistence 53 secret M_Keys 97 smnodes 95 Subnet Manager configuration integrity 54 controlled handover state 51 location history 51 log 55 management key 51 prefix 51 priority 51 routing algor
223. way on page 9 QSFP Connectors and Link LEDs to Switch Chip Port Routes Connector Connector Connector Connector Connector A Port Connector Group A Port B Port Group B Port 0 20 19 9 14 13 1 22 21 10 16 15 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Connector Connector Connector Connector Connector A Port Connector Group A Port B Port Group B Port 2 24 23 11 18 17 3 26 25 12 11 12 4 28 27 13 9 10 5 30 29 14 7 8 6 35 36 15 5 6 7 33 34 0 3 ETH 1 ETH 2 4 ETH 3 ETH 4 8 31 32 1 1 ETH 1 ETH 2 2 ETH 3 ETH 4 Related Information m Display the Switch Chip Port to QSFP Connector Mapping on page 32 m Display Link Status on page 35 m Display Switch Chip Port Status on page 37 m Switch Chip Port to QSFP Connectors and Link LED Routes on page 8 m Signal Route Through the Gateway on page 9 Signal Route Through the Gateway By combining the information from the tables in Understanding Routing Through the Gateway on page 7 it is possible to determine a route through the gateway This topic describes a sample situation that might occur 1 A route is initiated at connector 2A The LED blinks 2 Using QSFP Connectors and Link LEDs to Switch Chip Port Routes on page 8 it is determined that the link routes to the switch chip through port 24 3 The Subnet Manager instructs the switch chip to use port 5 to
224. work Management Troubleshooting Guidelines on page 6 Understanding Routing Through the Gateway on page 7 Switch GUIDs Overview on page 10 InfiniBand Fabric Problems This table lists situations that might occur with the InfiniBand fabric and corrective steps that can be taken to resolve the problem Situation Corrective Steps After installation no 1 Verify that there is at least one Subnet Manager active on the InfiniBand fabric links are operational After installation not all links are operational There was a power outage during a firmware update See Display Subnet Manager Priority Controlled Handover State Prefix Management Key and Routing Algorithm on page 51 If no Subnet Manager is active start the Subnet Manager within the gateway Refer to Gateway Installation staring the Subnet Manager If the previous steps do not rectify the situation restart the Subnet Manager See Disable the Subnet Manager on page 63 and Enable the Subnet Manager on page 62 Determine which links are nonoperational See Display Link Status on page 35 For links that are Down disable and re enable the respective ports See Disable a Switch Chip Port on page 40 and Enable a Switch Chip Port on page 41 If the previous steps do not rectify the situation disable the respective port See Disable a Switch Chip Port on page 40 If you are a
225. xm_main 1 31 mlx_bx_core 1 31 Parameter Configured Value Running Value GwiInstance 42 42 SystemName None o4nm2 gw 4 EoIB Data SL None 1 EoIB Control SL None 2 Allow host VNIC config None no LAG mode yes yes 182 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Default discover P_key None Oxfffft System MAC Not applicable 00 3b aa bb a0 15 FabMan gateway_name gt 2 If the information displayed is incorrect modify the values See m Set the Gateway Instance Number on page 207 m Set the Gateway System Name on page 208 m Set the Gateway Service Level on page 206 m Gateway Reference allowhostconfig command m Enable LAG Mode on page 184 m Disable LAG Mode on page 189 m Gateway Reference setdefaultdiscpkey command Related Information m Gateway Reference showgwconfig command a Display LAG Information on page 173 a Display the VLANs on page 174 m Display the VNICs on page 176 m Display IO Adapters on page 177 a Display Gateway Ethernet Port Information on page 178 a Display Gateway Port Information on page 179 a Display the Gateway Instance Number on page 181 Controlling LAGs LAGs are a means of effectively increasing the bandwidth of an Ethernet network by grouping ports m Enable LAG Mode on page 184 m Create LAGs on page 185 m Create a Global Gateway on pag
226. y See Display the InfiniBand Fabric Partition Configuration on page 93 3 Return to the gateway MAC address list and determine which addresses will be associated with a VLAN and which will be associated with VLAN ID No 4 For each MAC address append the VLAN identifier No or 2 to 4094 and partition key number to the right of the MAC address Note Do not use VLAN identifiers 0 1 or 4095 According to the IEEE 802 10 specification VLAN ID 0 is used for priority tag and VLAN ID 1 is usually reserved for a switch or bridge management VLAN Note Given a connector and vlan_id combination including NO there is only one p_key permitted Administering Gateway Resources 145 Note Due to hardware limitations for MultiCast groups there is a maximum of 1000 VLANs Even if the MAC address will not be associated with a VLAN VLAN identifier NO you must assign a partition key default or other Note When you associate a MAC with a VLAN the GUID respective to the MAC must be a member of the partition associated with the partition key 5 Use the list for when you create the VLANs and VNICs An example entry in the gateway MAC address list might look like this OA ETH 1 0003ba000100c70b 00 30 48 7d de e4 3 default For this example 0A ETH 1 is the gateway connector 0003ba000100c70b is the HCA port GUID 00 30 48 7d de e4 is the MAC address 3 is the VLAN identifier default is the
227. y 1 to 7 f or default Setting a discovery P_Key limits the Global Gateway s visibility m si is the data service level 0 to 15 and is used to discriminate between different classes of InfiniBand traffic 186 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Note Use service level 1 s1 1 only for low latency high priority small message low bandwidth traffic Use other service levels for regular high bandwidth traffic For example FabMan gateway_name gt createlag globalgate datasl 2 lag created FabMan gateway_name gt Related Information m Gateway Reference createlag command m Enable LAG Mode on page 184 m Disable LAG Mode on page 189 Add or Delete Connectors From a LAG You can add more connectors to a LAG to increase its bandwidth Or you can remove connectors from a LAG to free them up for other purposes Note A connector is assigned to only one LAG at a time Note If a VNIC has been assigned to a connector that connector cannot be added to a LAG until the respective VNIC has been deleted Note If the LAG is a Global Gateway you cannot add any connectors to the LAG 1 To add a connector to a LAG on the management controller type FabMan gateway_name gt addlagport lagname connector1 connector2 connectorN where m lagname is the identifier of the LAG m connector is the n
228. y value is effectively increased by 0x8000 Similarly if you were to define a P_Key with a value greater than 0x8000 the membership bit is automatically set to 1 and consequently is given full membership 108 Sun Network QDR InfiniBand Gateway Switch Administration Guide for Firmware Version 2 1 e March 2013 Related Information smpartition Command on page 106 smpartition Command Guidelines on page 107 Determine the Partitions and P_Keys on page 109 Create a User Partition on page 110 Add or Remove a Port From a Partition on page 112 Modify a Partition or Port on page 114 Delete a Partition on page 116 Remove User Partitions for Firmware Downgrade on page 117 V Determine the Partitions and P_Keys 1 Use the output of the ibswitches ibhosts and showgwports commands to identify your switch HCA and gateway node GUIDs See m Identify All Switches in the Fabric on page 84 m Identify All CAs in the Fabric on page 85 a Display Gateway Port Information on page 179 Determine the partitions you will have their names and their respective P_Keys Collate the GUIDs partition names and P_Keys into a partition information text file Optional If you will also configure VLANs assign a unique VLAN identifier to similar P_Keys in the text file Related Information Gateway Reference inswitches command Gateway Reference ibhosts command Gateway Reference

Download Pdf Manuals

image

Related Search

Related Contents

RT-4000 - RTM Injector Trigger Manual  Rheem Prestige Series: Up to 96% AFUE ECM Motor Variable Speed Multi Position Specification Sheet  FIA オリンピアクラス車両規則 2015  Maytag NEPTUNE MD68 User's Manual    - Hitachi Construction Machinery Europe  Primavera Analytics Licensing Information User Manual Release  Betriebsanleitung  Intenso Memory 2 Move MAX  Appendix - FTP Directory Listing  

Copyright © All rights reserved.
Failed to retrieve file