Migrating Symantec Veritas Cluster Server to HP Serviceguard for
Contents
1. cmviewcl CLUSTER STATUS apache cluster up NODE STATUS STATE Hnode up running Pnode up running GUI Serviceguard Manager Launch Serviceguard Manager from one of the nodes and click on the cluster name to view the cluster status The cluster status should be up Create and configure the Serviceguard Apache toolkit CLI Please follow the following steps to create and configure Serviceguard Apache toolkit Create Apache toolkit package directory mkdir SGCONF pkg Apache Change directory to the Apache toolkit package directory cd SGCONF pkg Apache Create the package configuration template using the cmmakepkg command and include appropriate modules required by your application In this illustration we can use the tkit apache module that is available through the SG LX Contributed Toolkit product cmmakepkg m tkit apache apache apache package conf cmmakepkg command by default includes all the modules necessary to manage typical application resources like filesystem volume group IP subnet etc in a package configuration file In addition you can select the application specific modules that can start stop monitor applications The application specific modules are available through corresponding toolkit product To get a list of modules installed in the cluster use the cmmakepkg l command If you do not find an appropriate module for your application buy or get the appropriate SG LX2. ice and logs messages in th on method which The is also ca node 1 and node 2 both nodes race to acquire the lock and node 2 wins the race by acquiring the allowed to continue running and node 1 is reset Figure 1 2 node SG LX cluster using Lock LUN as a cluster lock SG LX node 1 SG LX node 2 e syslog in case of a failure The Lock LUN is led as cluster lock HP recommends that the Lock LUN can be used only in the cluster which has nodes up to four The ollowing diagram shows Lock LUN configuration in a 2 node cluster When there is loss of communication between OCk The node 2 is Root Disk array Root rum DERE ER CEU CNN Hd om Ul wc no UA m m cum um na nico MEN ce LI we DE DN CONI Mirror Mirror Hub Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Quorum Server software Quorum Server can be used as amembership arbitration method for multiple clusters at a time The Quorum Server does not have any limitation in terms of number of nodes in a cluster The Quorum Server can be configured as a Serviceguard package or standalone but in either case it must run on a system outside of the cluster for which it is providing quorum services The Quorum Server listens to connection requests from the Serviceguard nodes on a known port The server maintains a special area in memory for each cluster and when a node obtains the clust
3. that simplifies the installation process It installs the Serviceguard components RPMs and list all prerequisite RPMs required for the Serviceguard components RPM installation Please refer the section Gather Required Software of SG LX Deployment Guide http bizsupport1 austin hp com bc docs support SupportManual c03393068 c03393068 pdf to get details about Serviceguard software installation Please refer sections Preparing the Servers Serviceguard System Configuration Automation script Allowing Root Access to an unconfigured Node and Configuring Name Resolution of SG LX Deployment Guide http bizsupport1 austin hp com bc docs support SupportManual c03393068 c03393068 pdf to get details about preparing nodes for the SG LX cluster The Apache application should not be reinstalled or reconfigured on the nodes The same Apache application installation and configuration used in the VCS cluster can be used in the SG LX cluster Create the shared storage for application data As in VCS cluster the application data that can change must be kept in a shared storage accessible from all the cluster nodes Application binaries and any application data that is smaller in size and don t change can be kept locally in each cluster node If the filesystem or volume manager used by the application to store its data in the VCS cluster is also supported by SG LX then the same storage layout can be used as is in an SG LX
4. A administrative privileges for the cluster strative privileges for DMIN plus 20 Parallel package type In VCS the Parallel attribute indicates if service group is failover 0 parallel 1 or hybrid 2 In Serviceguard the package type attribute is used to determine the type of package Following are the possible values of the package type attribute failover multi node system multi node 21 Priority priority service group and the sequence of actions r a particular event n Serviceguard the priority attribute is us highest priority equired in res ed n VCS the Priority attribute enables users to designate and prioritize the service group VCS does not interpret the va ather this attribute enables the user to configure the priority of a lue ponse to to specify priority of a package The lowest value of priority attribute mea ns the 22 SystemList node_name nVCS the SystemList attribute is used to heir priori system as a failover target systems on which the service group is configu es Lower numbers indicate a preference for the specify list of red to run and n Serviceguard multiple nodes can be specified as values for the node name attribute These multiple values serve as an ordered ist which is used to start or failover the package 23 UserlntGlobal pev
5. Administrator has the most privileges In SG LX you can also enable or restrict the uses by defining access control policy parameters Following three parameters are used to achieve this user name It can either be any user or a maximum of eight login names from the etc passwd file on user host user host It is where the user can issue Serviceguard commands It can be any serviceguard node or any cluster member node or a specific node user role t can have one of the following three values MONITOR read only capabilities for the cluster and packages PACKAGE ADMIN MONITOR plus administrative privileges for packages in the cluster FULL ADMIN MONITOR and PACKAGE ADMIN plus the administrative privileges for the cluster Mapping VCS system capacity and service group load to Serviceguard node capacity and package weight The VCS capacity construct enables you to define a fixed amount of resources a server can provide It sets a fixed load handling capacity for servers The Load attribute allows you to define a fixed amount of resources a specific service group is expected to utilize Load It sets a fixed demand for service groups When a service group is brought online its load is subtracted from the system s capacity to determine available capacity VCS maintains this info in the attribute Available Capacity When a failover occurs VCS determines which system has the highest available capacity and st
6. ERU A HS Summary 3 0 Critical W o Major 4X0 Minor CO 3 Normal 0 Disabled 1 unknown 0 informational Total 4 zi c Cluster Name Cluster Address Cluster Type Cluster Description n o ODE IIDCEEEDITI A F dee cluster 1511803849 HP Serviceguard HP Serviceguard Cluste t um s y a OPERATION boost 2012 09 12 23 12 520077826 HP Serviceguard HP Serviceguard Cluste perm la xylot_cluster_1707488373 HP Serviceguard HP Serviceguard Cluste ruClusters a Bi HP serviceguard Pron ESSE kz Fusion 1209 cluster 6 lee cluster 1511803849 l OPERATION boost 201 Bl xytot cluster 17074883 xylot xylo2 E Clusters by Status E Critical Clusters Major Clusters m cl m 17 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Figure 6 HP SMH Serviceguard Manager plug in showing a SG LX cluster HP Serviceguard Manager Cluster spicy Custer Status op Chster Aer O Node Alerts nos Package Alert Q2 Migh Availabilty A Alert A J Vew Administration Configuration Alerts loga System Tools vwws Mas Large Scsm Gra C 5 mmm lg ov IET BD onm Bre Grenn Vus IAM G 1 boe beei Eocay ag spicy L m Reporting IesIService Serviceguard command line interface Refresh Data 7 auto Fi n we See e a 9 Ton ga ga A a G Inventory Log WebAccess Serviceguard provides ric
7. MUN M DET 17 Serviceg ard command line Interface anite i re EH RU HERE KR ERE AERE G ee 18 Mapping of some important VCS commands to SG LX commands o ssessssesssessssessessssessscsssessscssscssecssecssecssecsseesseees 18 Migration procedure illustrated with a sample EXAMPLE essent tete tetettetnntnnens 19 Th nigraHuompFOCedUE eorr e EE E E AFER ERE 20 Plan SG LX heartbeat network and membership arbitration tentes 21 Collect the resources information used by your VCS service group teretes 22 Back up the application and configuration data eerte ttettettetthttite tette ttn tth ttn tth than 22 Install Serviceguard and Apache toolkit software and prepare the nodes 23 Create the shared storage for application data cccsccssssssscessnssscsssssscsssssosssssscsccscssesssssssssassscncassacassasesacacaeasensees 23 Configure the SG LX CLUStOr eeceessecsssesssesssesssssssscossessscssscsnsessscsssesssessscssscsssesssesssestsessseassesssesssssssessscstecsseessecssecsseesnecsseeseeess 23 STOP th VES CLUSEQT iicet tritt eh PRORFUS EL REA Hearst 24 SES Cables G X CLS be IR RE ROTER A 24 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Verify the SG LX cluster Create and configure the Serviceguard Apache toolkit wc cccecscesssssesessessessessesscsussessesseesessssussussesesseeseeseesesesateseeseesss 25 StartthieServiceguard Apache toolkit package nissana i e a
8. Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux 20 Figure 8 Target SG LX cluster running the highly available Apache Web server 192 168 4 244 Apache IP address Apache service groups SG LX heartbeat 10 25 1 1 10 25 1 2 REN SG LX heartbeat RES binaries binaries and config and config files files hnode pnode fo Soy ips Lock LUN mount vg_apache apache document root mounted using dev vg apache lvolO The migration procedure Overview This migration procedure can convert the existing VCS cluster to SG LX cluster using the same nodes and heartbeat network The target SG LX cluster can use a new shared disk as a cluster Lock LUN mechanism for membership arbitration The Apache web server configured in the service group apache svcgrp can be converted to an SG LX package with the name apache package using the SG LX contributed toolkit The sample Apache service group uses the following resource agents in the VCS cluster Apache P NIC VM volume group VM logical volume Mount Disk reservation In this migration procedure the same Apache application installation and configuration used in the VCS cluster is being used in the SG LX cluster The Apache Web server documents HTML files kept in a shared storage is also used as is in the SG LX package The Apache application is not reinstalled or reconfigured for this migration NOTE In ca
9. Operators user role OS user groups or users who have Operator privileges on the cluster n Serviceguard cluster and package access control policy parameters can be used ogether to get the different level of privileges for multiple users The user role parameter determines the role of the user Following are the supported values for user role attribute MONITOR read only capabilities for cluster and packages PACKAGE ADMIN MONITOR plus administrative privileges for packages FULL ADMIN MONITOR and PACKAGE ADMIN plus the administrative privileges for the cluster 11 PanicOnNoMem No attribute n VCS PanicOnNoMem indicates the action that you want VCS engine HAD to ake if it cannot receive messages from GAB due to low memory Serviceguard usually resets the system when sufficient memory is not allocatable 12 UseFence forced_pr_disable n VCS the UseFence attributes to enable or disable SCSI 3 persistent reservation PR based 1 0 fencing n Serviceguard SCSI 3 PR based I O fencing is enabled by default To disable SCSI 3 PR based I O fencing for all packages in the SG LX cluster set the parameter orced pr disable in the cluster configuration file to 1 Note HacliUserLevel HostMonLogLvl LogClusterUUID PrintMsg ProcessClass ProcessPriority SourceFile Stewards TypeLimit UserNames EngineShutdown DeleteOnlineResource CounterMissTolerance EngineClass CounterMissAction and Coun
10. VCS agents fall into three categories Bundled agents These agents are packaged with VCS These include agents for disk IP mount etc Enterprise agents The enterprise agents control third party applications These include agents for Oracle Sybase and Database 2 DB2 Custom agents Typically custom agents are developed because the user requires control of an application that the current bundled or enterprise agents do not support Usually customers or Symantec consultants develops the custom agents SG LX packages and toolkits overview SG LX packages The SG LX packages are the means by which Serviceguard starts and halts configured applications An SG LX package can be treated as a single unit of failover in the SG LX cluster The SG LX package can be classified into three groups Failover packages A failover package runs on one node at a time In case of a service network or other resource or dependency failure the package failover takes place A package failover involves both halting the existing package and starting the new instance of the package on a new node Figure 4 SG LX package moving during failover node 1 node 2 node 3 pkga E pkg Multi node packages A multi node package can run on multiple nodes simultaneously It can be independently halted or started on the individual nodes Failures of package components such as applications services generic resource or subnets may cause the package t
11. cluster also If the filesystem or volume manager used by the application to store data in the VCS cluster is not supported with SG LX then it is required to create a fresh storage layout for application data using filesystem or volume managers supported by SG LX For more recent list of SG LX supported filesystems and volume managers please refer to the HP SG LX Certification Matrix http bizsupport1 austin hp com bc docs support SupportManual c03398707 c033987 07 pdf Configure the SG LX cluster CLI Execute the following command to create the cluster ASCII file from either of the nodes as root user cmquerycl L dev sdf n hnod n pnod C cluster ascii Some important attributes of the cluster ascii file are listed below CLUSTER NAME hnode cluster NODE NAME hnode ETWORK INTERFACE ethO STATIONARY IP 92 168 4 230 ETWORK INTERFACE bondO HEARTBEAT IP 0 25 1 1 CLUSTER LOCK LUN dev sdf NODE_NAME pnode ETWORK INTERFACE ethO STATIONARY IP 192 168 4 231 ETWORK INTERFACE bondO HEARTBEAT IP 10 25 1 2 CLUSTER LOCK LUN dev sdf EMBER TIMEOUT 14000000 default value 23 Technical white paper Migrating Symantec Veritas Clus 24 The cmquerycl co a Quorum Server You can refer to c cluster behavior cmquerycl can au suffix cluster attribute in the ge Execute the fol cmapplyconf mmand networks for heartbeat T u
12. toolkit MySQL toolkit PostgresSQL toolkit Samba toolkit Sendmail toolkit Tomcat toolkit Serviceguard Developer s Toolbox Serviceguard Developer s Toolbox enables users to write a toolkit themselves Typically users write toolkits using Serviceguard Developer s Toolbox for that application whose toolkit is not available The Developer s Toolbox can also be downloaded from software hp com for free Mapping VCS service group attributes to SG LX package attributes This section compares VCS service group with SG LX package and describes the possible mapping of attributes VCS enables users to specify all hardware and software resources in a single service group VCS defines resource type for each resource it manages For example the IP resource type can be used to manage IP address resource VCS provides corresponding agent for each resource type A single resource agent manages all resources of that type For example the IP resource agent manages all IP resources Like VCS SG LX also enables users to specify the hardware and software resources in a single unit That single unit is called package in a SG LX cluster Users can specify services disk volume IP addresses generic resources etc in a single package SG LX also provides separate script to manage each resource These separate scripts are called modules and these can be specified while creating a package Examples of some modules are filesystem volume group and pa
13. toolkit for the application and install the same on all nodes If there are no toolkits available for your application you must write your own using SG LX Developer s Toolbox that is freely downloadable from software hp com Edit the package configuration file and configure the attributes using the VCS resource information and VCS agent information collected at the step Collect the resources information used by your VCS service group in Table 6 Package worksheet 25 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux 26 Use the following table to map VCS service group resource attributes to SG LX package attributes for this illustration VCS service Corresponding Descriptions group resource SG LX package attributes attributes Service group package_name Represent the name of service group package ConfigFile httpd config file Represent the path of Apache Web server configuration file P address ip address Represent the IP address used by the application Logical volume s name Represent the shared device name which is being used by the application Volume group Vg Represent the volume group name being used by the application ount S directory Represent the location where the filesystem being used by the application is mounted FS type s type Represent the type of the filesystem type being used by the application Using this
14. values dev sda dev sdb dev sdc Excluding the above Coordination point disks we can use a new disk dev sdf in the target SG LX cluster as cluster Lock LUN el Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux 22 Collect the resources information used by your VCS service group The resource information of the VCS service group for your application can be used later while creating an SG LX package for the same application in the SG LX cluster The service group resource information typically includes storage filesystems volume managers network IP subnets and other OS resources used by the application Collect the following details related to the IP being used by the application IP device IP address Netmask NIC device Collect the following detail related to the volume group being used by the application Volume group Collect the following details related to the logical volume group being used by the application Logical volume and corresponding volume group Collect the following details related to filesystem being used by the application MountPoint FSType FsckOpt BlockDevice In addition to the resource information there can be VCS application agent specific information in the VCS service group that can also be useful when configuring SG LX package using an SG LX toolkit This typically includes the follow
15. whether a service of package can be restarted or not 4 AutoStart auto run n VCS the AutoStart attribute designates whether a service group is automatically started when VCS is started n Serviceguard the auto run attribute determines whether a service group is automatically started when VCS is started or not 5 AutoStartList node name n VCS the AutoStartList attribute is used to specify the list of systems on which under specific conditions the service group can be started with VCS usually at system boot n Serviceguard the node name attribute is used to specify the order of nodes in which package can start 13 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux VCS service group Corresponding SG LX package attributes attributes which can be equivalent to some extent present in editable by users package ASCII file which can be editable by users Descriptions 6 AutoStartPolicy failover_policy In VCS the AutoStartPolicy attribute sets the policy which VCS uses to select the system to bring a service group online if multiple systems are available This attribute has three options Order Load and Priority In Serviceguard the following failover policies can be used to determine which node should be chosen to bring up a package configured node ordered list of nodes min package node A node which has minimu
16. Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Table of contents EXecultive SUMMAN seis eot teni cet era iacta rei hen Det noe tas erbe d EE egestas ERR be REB RR AREE Ig ERE ERR e es 3 Joeersi fenfoee LEE 3 pus 3 Convertirig a VES Setup tO SG LX Sebp ette e t t ERO ER RR RR REDEEM 3 SB NO EIV AO 3 ries SM WV c 6 Comparison of some important terminologies sese ttettettntthte tette teta ariii 7 Mapping of VCS cluster attributes to SG LX cluster attributes tenentes 7 Mapping of VCS system attributes to SG LX node attributes sssssetetettette tenetis 9 Converting VCS service groups to SG LX packag Suesscssssssrsnonsnsnsninininikini kinikini kanhi RRASA 9 The VCS service groups and the SG LX packages tette tette teni tentent teen 9 VCS service groups and agents OVELVICW eceecsesessessessessessessecsessssussucsussessessessesussucsucsssssessessesesussussueseseessarsausaucsteseeseeseeses 10 SGX packages and toolkits OVerVIBWw ctr rtr ttr rette ip nt peperere ap oe oe 10 Mapping VCS service group attributes to SG LX package attributes 11 Mapping VCS service group dependency to Serviceguard package dependency 11 Administering an SG LX cluster for a VCS administrator tentenitenttetienin 17 SenvieegdakdiMariedel il ecce eorura a AM A AM AMT M
17. OnlineRetryLimit service restart In VCS the OnlineRet yLimit attribute specifies the number of times a VCS engine tries to restart a faulted service group on the same system on which the service group faulted before it gives up and tries to fail over the group to another system In Serviceguard the service restart attribute specifies the number of time the service of the package can be restarted on a particular node 15 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux VCS service group attributes which can be editable by users Corresponding SG LX package attributes equivalent to some extent present in package ASCII file which can be editable by users Descriptions OperatorGroups Operators user_name user_host and user_role unning in secure mode The user_role parameter determines the ro n Serviceguard cluster and package access control policy parameters are used to specify multiple users with diffe le of the user Following are the supported values for user_role attribu nVCS the OperatorGroups and Operators are used to specify a list of OS user groups and VCS users who can have Operator privilege on the service group This attribute applies to clusters ent roles e MONITOR read only capabilities for cluster and packages PACKAGE ADMIN MONITOR plus admini packages FULL ADMIN MONITOR and PACKAGE
18. ach other at different data centers over metropolitan distance This solution integrates SG LX robust failover mechanism with disk array based replication technologies This allows applications to automatically failover across data centers and use disk array based replication between the data centers The SG LX failover can automatically manage the replication preserving data integrity and continuing protection against future failures Metrocluster for Linux is available for the following HP array based replication technologies HP 3PAR P10000 P7000 Remote Copy HP XP P9000 Continuous Access and EVA P6000 Continuous Access Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux HP Continentalclusters for Linux HP Continentalclusters protects and manages multiple clusters where the clusters are located apart from each other over any distance With Continentalc on the recovery site and use any type of replication to replicate data be usters one can perform a push button recovery tween the data centers HP provides preintegrated solutions with Continentalclusters and disk array based replication for the following arrays like HP 3PAR P10000 P7000 Remote Copy HP XP P9000 Continuous Access and EVA P6000 Contin uous Access In addition you can integrate any host based replication technology with Continentalclusters using the data sender and data receiver framework in Continentalclusters You can also
19. acuate cmhaltnode f n VCS the Evacuate a before stopping VCS on n Serviceguard the cm behavior tribute initiates an automatic failover he node haltnode f command initiates automatic ailover of the packages provided the global and node switching is enabled for those packages This is a default Serviceguard 10 FailOverPolicy failover policy ollowing values n VCS the FailOverPolicy attribute sets the policy which VCS uses to select a system on which a service group will failover o if multiple systems exist This attribute can take the Priority The system defined as the lowest priority in the SystemList attribute is chosen Load The system defined with the least value in the system s Load attribute is chosen Roundrobin Systems are chosen according to how many active service groups they are hosting The system with the least number of active service groups is chosen first In Serviceguard the failover policy attribute determines which node be chosen to start or failover the package Following values are used for failover policy parameter in Serviceguard cluster configured node The node which is first in the eligible nodes list is selected as an adoptive node min package node The node which is running the minimum number of packages is selected as an adoptive node 14 Technical white paper Migrating Symantec Veritas Cluster Serve
20. antec Veritas Cluster Server to HP Serviceguard for Linux 12 Following are the possible values for dependency location in VCS Local dependency The parent service group depends on the child service group for being online or offline on the same system Global dependency An instance of the parent service group depends on one or more instances of the child service group being online on any system in the cluster Remote dependency An instance of the parent service group depends on one or more instances of the child service group being online on any system in the cluster other than the system on which the parent service group is online Following are the corresponding possible values for dependency location in SG LX same node It means that the dependency must be satisfied on the same node any node It means that the dependency can be satisfied on any node in the cluster different node t means that the dependency must be satisfied on a node other than the dependent package s node Following are the possible values for the dependency rigidity in VCS Soft dependency It specifies the minimum constraints while bringing parent and child groups online Firm dependency It imposes more constraints when VCS brings the parent or child groups online or takes them offline Hard dependency It imposes the maximum constraints when VCS brings the parent of child service groups online or takes them offline There is no con
21. approach the SG LX package configuration file in this illustration can be populated as below SG LX package attributes package name apache package You can use the same VCS service group name also kit dir SGCONF pkg Apache httpd config file etc httpd conf httpd conf ip_subnet 192 168 0 0 ip address 192 168 4 244 vg vg_apache s_name dev vg_apache lvol0 s_directory mount vg_apache apache s_type ext3 Apply the package configuration cmapplyconf P apache_package conf GUI Serviceguard Manager Please refer the sections Create Package from Serviceguard Manager and Customize and Distribute Toolkit Files of HP SG LX Deployment Guide http bizsupport1 austin hp com bc docs support SupportManual c03393068 c03393068 pdf to get details on using Serviceguard Manager to configure Apache toolkit package Please refer to the preceding table while configuring the Apache toolkit package Please refer section Apache Toolkit of HP Serviceguard Contributed Toolkit Suite on Linux User Guide http bizsupport1 austin hp com bc docs support SupportManual c03603547 c03603547 pdf to get more details of Apache toolkit configuration Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Start the Serviceguard Apache toolkit package CLI Execute following command to start the cluster cmrunpkg apache_package GUI Serviceguard Manag
22. arranties for HP products and services are set forth in the express warranty statements accompanying such products and services Nothing herein should be construed as constituting an additional warranty HP shall not be liable for technical or editorial errors or omissions contained herein Oracle is a registered trademark of Oracle and or its affiliates 4AA4 5152ENW March 2013
23. arts the service group on that system Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux SG LX also enables you to define the capacities for a node and corresponding weight for a package The capacity of a node determines how much weight of packages it can accommodate For example you can define a capacity called memory which has value 1000 for a node This means that node can accommodate a package or multiple packages having weight less than or equal to 1000 If you have defined package A with weight 1000 package B with weight 500 and package C with weight 500 then that node can either accommodate only package A or both package B and package C In SG LX you can use following parameters to define capacity of nodes and weight of packages You can define multiple capacities and weights if required capacity name It provides the name of the capacity This is a node attribute capacity value It provides the value of the capacity This is a node attribute weight name It provides the name of the weight It should be the same as one of the capacity names This is a package attribute weight value It provides the value of the weight This attribute is associated with the package The value of weight gets subtracted from the node s capacity value where it is running SG LX uses first fit policy when capacities and weights are configured Based on the failover policy of a package SG LX cre
24. ates a list of adoptive nodes for failing over that package If capacities and weight are configured SG LX chooses the first adoptive node which can accommodate the package weight for failover Table 4 VCS service group attributes and SG LX package attributes VCS service group Corresponding SG LX package attributes Descriptions attributes which can be equivalent to some extent present in editable by users package ASCII file which can be editable by users 1 AdministratorGroups user_name user_host and user_role nVCS the AdministratorGroups and Administrators attributes Administrators are used to provide administrative privileges on the service group o list of OS user account groups and users Serviceguard provides access control policy parameters to provide Administrator access to multiple OS users Administrator access o multiple users is provided by listing multiple users with FULL_ADMIN as USER_ROLE 2 AutoFailOver auto_run n VCS the AutoFailOver attribute indicates whether VCS initiates an automatic failover if the service group faults n Serviceguard the auto run attribute is enabled to initiate an automatic failover of a package if that package fails 3 AutoRestart Service restart n VCS the AutoRestart attribute is used to restart a service group after a faulted persistent resource becomes online n Serviceguard the service restart attribute is used to determine
25. can be altered rmine which nodes get to fence off data of communications intra nter system co uster systems The Group uster commun is powered off it stops se 21 seconds before declar ncy Transport LLT peer i runs a kernel fencing mod bership change through t ion point for membership by users e with software to provide application failover and control When a node or a monitored application n take predefined actions to take over and bring up the failed application elsewhere in the cluster system communications and inter system communications In the intra mmunications VCS uses the cluster interconnects for network Membership Services Atomic Broadcast protocol GAB is responsi ications e nding heartbeats to other systems in the cluster By default other ing it dead The time of 21 seconds derives from 16 seconds defau nactive timeout can be modified by users plus 5 seconds default ule which is responsible for ensuring valid and current cluster he process of membership arbitration VCS cluster uses a special arbitration Coordination points provide a lock mechanism to drives f rive The r from the data d rdinator points The kernel fencing modu ation At the time of cluster reformation the kernel fencing module of each system races for the control of the igured coordination points VCS prevents split brain by allowing the winner partition to fence the ejected nodes from rom other nod
26. ckage ip Mapping VCS service group dependency to Serviceguard package dependency A VCS service group can be dependent on another service group Same service group may act as both parent and child in a dependency tree In VCS a dependency tree may be up to five levels deep The dependency relationship between a parent and a child service group can be characterized by the dependency category the location of the service groups and the rigidity of dependency Similarly an SG LX package can be also dependent on another package The same package can act as both parent and child in a dependency tree The SG LX dependency relationship between a parent and a child can be characterized by the dependency name condition and location of the package Following are the possible values for dependency categories in VCS Online group dependency The parent service group must wait for the child service group to be brought online before it can start Offline group dependency The parent service group can be started only if the child service group is offline and vice versa Following are the corresponding possible values for the dependency conditions in SG LX up The dependent successor package must wait for the predecessor package to be brought online before it can start down The dependent successor package can be started only if the predecessor package is offline and vice versa 11 Technical white paper Migrating Sym
27. eated n required quorum is general number of nodes configured in the cluste Split brain If a failure in the SG LX clus er results in equal halves exactly 5096 of a previously running c nthe o refo thec he quo uster as one vote During rm a cluster In SG LX ition is met only when ster All and any rum is dynamic The atic count of the u S nultiple cluster node failures hen this situation is called as split brain A robust arbitration method is used to break a tie This robust arbitration method guarantees that only one half can form a cluster and the other half may be forced to shut down In SG LX arbitration can be implemented by cluster lock Either Lock logical unit nu mber LUN or Quo membership arbitration method cluster Lock LUN all nodes in the cluster During split brain Once the lock is obtained it is marked so Lock LUN should comprise of whole disk Lock LUN acts as a quorum arbitration method in the SG LX clust periodically checks the status of Lock LUN dev dedicated to use as only membership arbitrati OCk is a must in the clu ition tion situa that tion each par he other parti sters tha have two nodes rum Server can be configured as cluster lock Configuring a ers It is a special area in the storage which is sharable by races wi h the other to obtain a ock on the Lock LUN can recognize the Lock LUN as taken Serviceguard
28. er Launch Serviceguard Manager from one of the nodes Select the apache package go to the Administration menu and select Run Package Verify the Apache toolkit package CLI Execute the following command to view the status of the cluster cmviewcl CLUSTER STATUS apache cluster up NODE STATUS STATE Hnode up running PACKAGE STATUS STATE AUTO RUN NODE apache package up running enabled hnode NODE STATUS STATE Pnode up running Test that you can access the apache package content from your browser at the following URL http 192 168 4 244 80 index html Try halting the package with the following command cmhaltpkg n hnode apache package Try starting the package on other node with the following command cmrunpkg n pnode apache package Test that you can access the apache package content from your browser at the following URL http 192 168 4 244 80 index html GUI Serviceguard Manager Please refer the Verification section of HP SG LX Deployment Guide http bizsupport1 austin hp com bc docs support SupportManual c03393068 c03 393068 pdf to get the details on how to verify the apache package using Serviceguard Manager Remove the VCS configuration and software After verifying the Apache toolkit package you can remove the VCS configuration and VCS software Please refer to Uninstalling VCS section of Veritas Cluster Server Installa
29. er node and quickly responding to failures of system processes system memory LAN media and adapters or applications In the event when one component fails the redundant component takes over without compromising data integrity HP SG LX on its own does not require planned downtime for maintenance It can also help you perform application OS and hardware maintenance operations effectively against the ever shrinking planned downtimes available for such maintenance operations HP provides multiple extensions and toolkits that provide out of the box deployment of popular mission critical applications in a SG LX cluster The SG LX portfolio also contains other value added solutions that provide disaster recovery capabilities by leveraging robust Serviceguard automatic failover with data redundancy technologies Serviceguard Manager Serviceguard Manager is a GUI to administer SG LX cluster Extended Distance Cluster XDC The XDC protects against multiple failures by extending a single SG LX cluster across two data centers separated up to 100 km distance It uses host based data replication Toolkits HP Serviceguard provides toolkits to manage popular applications for example Oracle NFS SAP Apache and others HP Metrocluster for Linux HP Metrocluster is a disaster recovery solution that helps ensure service availability in the event of disaster It enables you to create a single cluster using nodes that are located apart from e
30. er lock this area is marked so that other nodes can recognize the lock as taken The operation of the Quorum Server in a 2 node cluster is shown in the following diagram figure 2 When there is a loss of communication between node 1 and node 2 the Quorum Server chooses one node in this example node 2 to continue running in the cluster The other node node 1 is reset Figure 2 2 node SG LX cluster using Quorum Server as a cluster lock Storage Storage SG LX node 1 SG LX node 2 NICT NIC2 NIC1 NIC Z h Quorum server amp c NICI NIC2 ae The following diagram shows a Quorum Server providing quorum service to multiple clusters Figure 3 Single Quorum Server providing quorum service to multiple clusters Standalone quorum server Storage Storage SG LX node 1 SG LX node 2 SG LX no a Ls g Cluster 1 m for cluster 1 2 3 and 4 Storage Storage de 1 SG LX node 2 al ai g g Cluster 2 _ node 2 SG LX no SG LX nod pr e 1 Storage Storage Cluster 3 9 9 de 1 SG LX node 2 Storage Storage Cluster 4 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux The Quorum Server software is available for HP UX 11i v2 v3 RHEL and SUSE Linux Enterprise Server The software can be downloaded free from software hp com A single Quorum Server can serve multiple SG LX and Serviceguard for HP UX clusters Idea
31. es A node must eject a peer from the coordination point coordination point can be a disk or a server VCS recommends e registers with the coordination points during normal Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Comparison of some important terminologies The following table compares the important VCS terminologies with SG LX terminologies Table 1 Important terminologies VCS terminology SG LX terminology Descriptions 1 System Node Member of cluster 2 Service group Package Way to package all hardware and software resource together as single unit 3 Resource agent Modules Way to manage start stop monitor hardware and software resources via SG LX package or VCS service groups 4 Agent Toolkit A framework to manage start stop monitor a specific application 5 Disk fencing Lock LUN One of the cluster membership arbitration mechanisms 6 Coordination point server Quorum Server One of the cluster membership arbitration mechanisms 7 VCS Operation Manager Serviceguard Manager GUI 8 Failover service groups Failover package A packaged single unit which contains software and hardware resources that run on one node system at a time 9 Parallel service groups Multi node package A service group package that can run on multiple nodes at a time 10 Global clusters Continentalclusters for Linux Disaster recovery solution in which multiple clusters are used t
32. f cluster and packages Serviceguard Manager is a browser based management tool available as HP System Management Homepage HP SMH plug in HP Serviceguard Manager plug in for HP SMH can be launched from any of the nodes of a HP SG LX cluster with a supported web browser It is installed and automatically configured as part of SG LX installation The HP System Insight Manager HP SIM can be used to manage multiple SG LX clusters The HP SIM shows status and events from multiple SG LX clusters and you can launch HP Serviceguard Manager plug in from HP SIM for any specific cluster Please refer the following diagram to see the HP SMH plug in launched via HP SIM Figure 5 HP SIM showing multiple SG LX clusters HP Systems Insight Manager Windows Internet Explorer File Edit View Favorites Tools Help x umm SORS Search FHis 960 m iiaa Rose GJ 9 0o M x sip Favorites xig Howto enable IP Forward http h20000 www2 hp co Web Slice Gallery e HP Systems Insight Manager Ap HP Systems Insight Manager i Tools Deploy Configure Diagnose Optimize Reports Tasks amp Logs v Options Help Debug v HP Serviceguard Maximize E Updated Friday 9 14 2012 2 24 PM IST eg ovaAO Systems SESW oi Launch 23 21 10 81 Uncleared Event Status Cust table v ustomize Search mtn Select HP Serviceguard itself
33. figurable parameter in SG LX cluster to define the rigidity of a dependency The VCS firm dependency rigidity is the closest equivalent for SG LX package dependency rigidity Mapping VCS resources to Serviceguard modules n VCS you can include resources in a service group and use corresponding resource agent for monitoring For example IP resource agent filesystem resource agent mount resource agent and others You can define the order in which the resources can start and stop by defining dependency among resources n Serviceguard you can also include resources in a package by including corresponding resource module For example modules of volume group IP filesystem and others These modules only start and stop the resources For monitoring the resources you can configure generic resources in the SG LX package To configure a generic resource a monitoring script needs to be written which can monitor the resource and set the status of generic resource SG LX recommends launching the monitoring scripts by configuring them as services in a package Mapping VCS user privileges and roles to SG LX access control policy parameters Cluster operations are enabled or restricted depending on the privileges with which you log on VCS has three privilege levels Administrator Operator and Guest VCS provides some predefined user roles each role has specific privilege levels For example the role Guest has the fewest privileges and the role Cluster
34. h set of commands to administer cluster and packages The following table compares some of the important VCS command with SG LX commands Mapping of some important VCS commands to SG LX commands Table 5 Some important commands of VCS and SG LX VCS command Description Corresponding SG LX command 1 Hastart To start the cluster cmruncl 2 Hastop To halt the cluster cmhaltcl 3 hagrp online service group sys system To bring up a service group package cmrunpkg n node name package 4 hagrp offline service group sys system To halt a service group package cmhaltpkg n node name package 5 hastatus sum To view the cluster status Cmviewcl 6 haconf makerw To add create service group or package cmmakepkg package configuration file nag p add service group haconf dump makero fill the package configuration file cmapplyconf p package configuration file 18 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux VCS command Description Corresponding SG LX command 7 haconf makerw To modify service group or package cmgetconf p package gt hagrp modify service group package configuration file haconf dump makero Edit the package configuration file as required cmapplyconf p package configuration file 6 haconf makerw To delete service group package cmdeleteconf p package hagr
35. ie ut if you are planning to update Linux kernel then it is required to heartbeat among all the nodes and to maintain the cluster for communication between nodes To build a resilient cluster it is ure So it is recommended to configure multiple heartbeat networks own network in the cluster for their client access Such networks are manage on the cli P addresses used by applications on these networks During ent access network from the failed node to the target node beat network traffic SG LX exchanges only small messages nt access networks for heartbeat does not affect rom Symantec VCS VCS connects multiple independent systems into a management framework for increased availability Each system or node runs its own OS and cooperates at the software level to form a cluster VCS links commodity hardwar fails other nodes ca VCS cluster membership VCS cluster uses two types system communications VCS uses protocol called Inter Process Messaging IPM to communicate with the GUI the ne and the agents In the i command li communica for cluster ion between cl em crashes or the cluster wait ue for Low Late AB stable ti When a sys systems in timeout va value for G m in the cluster ip during a mem ed the coordina Each syste membersh service cal dete befo conf oper conf accessing the data disks re it can fence the pee iguring three coo membership and reliable c meout
36. ing Collect the following details related to the application Application configuration file path Application home or binaries dir Application startup and shutdown behavior flags In this illustration the following table has the VCS Apache service group resource information and the VCS Apache agent specific information that can be required while creating an SG LX package for the Apache Web Server Table 6 Package worksheet Attributes Values Service group apache svcgrp You can use same name for SG LX package ConfigFile etc httpd conf httpd conf P device ethO P address 192 168 4 244 etmask 255 255 254 0 IC device ethO Logical volume dev vg_apache lvolO Volume group vg apache Mount mount vg_apache apache FS_type ext3 Back up the application and configuration data You should always back up the application data when you are going to make significant changes to a system configuration It is recommended to create a backup disk with all the application data Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Install Serviceguard and Apache toolkit software and prepare the nodes Install Serviceguard and Apache toolkit software on each node of the cluster The Apache toolkit software is a part of the Serviceguard Contributed Toolkit which is free of cost Serviceguard provides a utility cmeasyinstall
37. lly the Quorum Server must be configured in a separate service cluster for HA in fault domain different from the fault domain of servers in the cluster serviced by them Node fencing A reboot is done if a cluster node cannot communicate with the majority of cluster members for a predetermined time split initiated by DEADMAN dr brain or under other circumstances such as a iver which act as a robust kernel hang or failure of the cluster daemon cmcld This reboot is fencing mechanism in SG LX cluster There is no manual configuration needed to configure DEADMAN driver The DEADMAN driver is a dynamically loadable kernel module that is compiled into the kernel automatically reco Netwo SG LX uses one or more he membership SG LX also uses crucial to have resilient hea and use channel bonding Applications deployed in SG LX clu called client access networks SG he appl failover SG LX can move Also it is okay to configure and does not have demand application accessibility o VCS overview VCS is a HA solution mpile the kernel with D ech when SG LX is installed EADMAN driver X rking in the SG LX cluster beat networks to send the heartbeat network eat network infrastruc nologies at link level art rtb et ster can use their LX can monitor and ication s IP address the client access network ing bandwidth requirem r performance S for hear ents So using cl
38. m number of packages can be selected to bring up the package or failover the package 7 ClusterFailOverPolicy failover policy In VCS the ClusterFailOverPolicy attri bute determines how a global service group behaves when a cluster faults or when a global group faults The attribute can take the following values Manual The group does not fail over to another cluster automatically Auto The group fails over to another cluster automatically if it is unable to failover within the local cluster or if the entire cluster faults Connected The group fails over automatically to another cluster only if it is una With HP Metrocluster ac ble to fail over within the local cluster uster can be divided into sites The sites can be located far from each other Serviceguard provides different failover policies site preferred manu for the sites al This failover policy helps ensure that the SG LX package never fails over to another site automatically site preferred This ailover policy helps ensure that the SG LX package fails over to another site only when it is unable to fail over within the current site 8 Enabled auto run n VCS the Enabled a n Serviceguard the au can be failed over or bro ribute indicates if a service group can be ailed over or brought online 0 run attribute determines if a package ught online 9 Ev
39. n SITE NAME n VCS the SystemLocation attribute can be used to indicate the ocation of the system n Serviceguard SITE NAME attribute is used to indicate the ocation of nodes 5 Userlnt PEV_ n VCS the attribute Userlnt is used to store integer values VCS does not interpret the value of this attribute n Serviceguard user defined variables can specify using PEV prefix in the application s package configuration file Converting VCS service groups to SG LX packages The VCS service groups and the SG LX packages VCS service group acts as a virtual container that contains all the hardware and software resources that are required to run an application The VCS service groups allow VCS to control all the hardware and software resources of the managed application as a single unit The SG LX package can be defined as a collection of services disk volumes generic resources and IP addresses that are managed by Serviceguard to help ensure they are available In case of any hardware or software failure which impacts any component of the SG LX package Serviceguard fails over the package to another node in the cluster This allows the application services to continue with minimal interruption Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux VCS service groups and agents overview VCS service groups VCS service group can be defined as a set of resources working together to provide a
40. o be halted only on the node on which the failure occurred System multi node packages System multi node packages are supported only for the applications supplied by HP A system multi node package runs on all cluster nodes at the same time It cannot be started and halted on the individual nodes 10 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux SG LX toolkits The SG LX toolkits enable easy integration of applications with SG LX to provide HA SG LX provides separate toolkits for supported applications to create application specific Serviceguard package The SG LX toolkit starts monitors and stops the application SG LX provides the following toolkits Oracle database DB toolkit The Serviceguard Oracle DB toolkit manages Oracle DB It starts monitors and stops Oracle DB instance This toolkit is priced NFStoolkit The Serviceguard NFS toolkit manages NFS server It starts monitors and stops the NFS Server This toolkit can be downloaded for free from software hp com SGeSAP toolkit The Serviceguard extension for SAP toolkit manages SAP instance It starts monitors and stops the SAP instance This toolkit is priced HPServiceguard Contributed Toolkits The Contributed Toolkit is a collection of toolkits that integrate popular applications with SG LX Following are the names of Contributed Toolkits These toolkits can be downloaded for free from software hp com Apache
41. o provide application recovery over LAN or WAN 11 Campus or metropolitan XDC and Metrocluster A single cluster that spans across multiple data centers or buildings clusters for Linux 12 Cluster Cluster Group of servers Mapping of VCS cluster attributes to SG LX cluster attributes The following table maps the VCS cluster attributes to the closest SG LX cluster attributes Table 2 VCS cluster attributes and SG LX cluster attributes VCS cluster attributes which can be configured by users Corresponding SG LX cluster attributes equivalent to some extent which can be configured by users Descriptions AdministratorGroups Administrator user_name user_host and user_role In VCS the AdministratorGroups and Administrator attributes are used to specify list of OS user account groups or list of OS users who have administrative privileges on the cluster This attribute applies to clusters running in secure mode Serviceguard provides role based access control RBAC which is used to grant Administrator privilege to multiple OS users by specifying USER_ROLE as FULL ADMIN 3 ClusterName cluster name The name of the cluster 4 EnableFFDC cmcluster conf file n VCS you can enable or disable Fast failure data capture FFDC logging using EnableFFDC attribute By default FFDC logging is enabled n Serviceguard the etc cmcluster conf file is used to enable logging of various daemons 5 EnginePriori
42. o send emails on the package failure events Guests user name user host and user role USER ROLE n VCS the Guests at ribute is used to provide Guest privileges to the list of OS user on the service group n Serviceguard access control policy parameters are used to grant read only privilege to users by specifying MONITOR as Load weight name and weight value n VCS the Load attribute specifies an integer value expressing Otal system load the service group can put on a system n Serviceguard the weight name and weight value attributes are used to define the weight of an SG LX package The value of weight name shou be the same as one of the capacities defined in the cluster The value of a package s weight gets subtracted from the runs on that node node s capacity value where the package 16 ManageFaults No attribute In VCS the ManageFaults attribute specifies if VCS manages resource failures wit hin the service group by calling the clean function for the resources In Serviceguard on t needed to do a prope ailure of package manual intervention is cleanup 17 ManualOps No attribute In VCS the ManualOps attribute indicates if manual operations are allowed on the se Serviceguard does no on packages vice group recommend performing manual operation 18
43. p delete service group configuration haconf dump makero 7 haagent start agent name sys To start agent toolkit package cmrunpkg n node name toolkit package Migration procedure illustrated with a sample example This section describes a step by step procedure to migrate a VCS cluster running Apache Web server to an SG LX cluster The VCS cluster has two systems hnode and pnode running Red Hat 6 1 and uses three SCSI 3 based coordination point disks as membership arbitration mechanism There is a single channel bonded heartbeat network 10 25 1 0 between the nodes The Apache Web server application is deployed as a failover service group apache svcgrp using VCS Apache agent The Apache Web server is installed and configured locally on both nodes of the cluster that is the Apache configuration file process ID file and the binary that reside locally on both the nodes of the cluster The Apache document root mount vg apache apache resides on a shared disk dev vgapache lvolO accessible from both the nodes Figure 7 Symantec VCS cluster running the highly available Apache Web server M 192 168 4 244 Apache IP address Apache pias GAB LLT bondO Apache GAB LLT Apache binaries binaries and config and config files files hnode pnode e dev sda dev sdb dev sdc Coordination point disks mount vg_apache apache document root mounted using dev vg_apache lvol0 19 Technical white paper
44. pplication services to clients It can be made dependent on or independent of another service group In the event of a failure the entire service group fails over without affecting others if that service group does not have any dependency on another The service group dependencies can be used to create more complex managed applications A single node can host any number of service groups It can be classified into three groups failover parallel and hybrid Failover group A failover service group runs on one system in the cluster at a time Failover groups are used for most applications that do not support multiple systems to simultaneously access the application s data Parallel group A parallel service group is more complex than a failover group Parallel service groups are appropriate for the applications that manage multiple application instances that run simultaneously without data corruption Hybrid group A hybrid service group is for the replicated data clusters and is a combination of the failover and parallel service groups It behaves as a failover group within a system zone and a parallel group across system zones VCS agents VCS agents are multi threaded processes that manage resources VCS has one resource agent per resource type A single agent manages all resources of that type The agent starts the resource stops the resource periodically monitors the resource and updates the VCS engine with the resource status The
45. prefix is not used by VCS pev prefix VCS users can use the UserlntGlobal attribute for any purpose It n Serviceguard user defined variables can be specified using 24 UserStrGlobal pev prefix attribute for any purpose in other service gr n Serviceguard user defined variables can pev prefix OUDS be specified u VCS uses the UserStrGlobal attribute in the ClusterService group Do not modify this attribute in the ClusterService group Use the sing 25 UserlntLocal pev prefix VCS users can use the UserlntLocal attribu is not used by VCS n Serviceguard user defined variables can pev prefix te for any pu be specified u pose It sing 26 UserStrLocal pev prefix is not used by VCS pev prefix VCS users can use the UserStrLocal attribute for any pu pose It n Serviceguard user defined variables can be specified using 16 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux NOTE AutoStartlfPartial OnlineRetryInterval Prerequisites PreSwitch SysDownPolicy PrintTree and SystemZones are not relevant in SG LX environment Administering an SG LX cluster for a VCS administrator Serviceguard Manager GUI SG LX cluster can be administered by GUI The Serviceguard GUI is called Serviceguard Manager which simplifies the configuration and administration o
46. r to HP Serviceguard for Linux VCS service group attributes which can be editable by users Corresponding SG LX package attributes equivalent to some extent present in package ASCII file which can be editable by users Descriptions FaultPropagation No attribute n VCS the FaultPropagation attribute specifies if VCS should propagate the fault up service group offline when a resource faults 0 parent resources and take the entire n Serviceguard failure of any configured resource in a package can cause the package to failover GroupOwner email id n Serviceguard the A mail to the configured n VCS the GroupOwner attribute is used for VCS email notification and logging VCS sends email notification to the person designated in t elated to the service group Note that while VCS logs most events not all events trigger notifications his attribute when events occur that are ert notification feature is used to send email aliases on the package failure events Cc GroupRecipients email id n VCS the GroupReci n Serviceguard the A mail to the configured pients attribute is used for VCS email notification VCS sends email notification to persons designated in his attribute when events related to the service group occur and when the event s seve evel specified in the a ity level is equal to or greater than the tribute ert notification feature is used t
47. s fully automated and ready to support your multi platform environment and diverse workloads It helps maximize uptime by moving workloads between servers in the event of a failure so quickly and transparently that users may not even realize it happened The first half of this paper provides VCS overview SG LX overview mapping of a VCS cluster to an SG LX cluster mapping of a VCS system to an SG LX node mapping of a VCS service group to an SG LX package The second half of this paper describes a step by step procedure to migrate a VCS 6 0 cluster to an SG LX A 11 20 00 cluster by using a sample example Audience This document is targeted for the users of Symantec VCS who want to migrate to HP SG LX It is assumed that the reader has a general understanding of Symantec VCS and HP SG LX For more information on each solution see hp com go sglx and Veritas Cluster Server Administrator s Guide Converting a VCS setup to SG LX setup SG LX overview HP Serviceguard solutions for Linux A 11 20 00 provides critical applications with the HA that enterprise customers require for 24x7 business operations on an x86 platform SG LX allows users to create HA clusters of HP ProLiant servers that are running Red Hat Enterprise Linux RHEL or SUSE Linux Enterprise Server HP Serviceguard solutions for Linux HA clusters are designed to protect applications from a wide variety of software and hardware failures by monitoring the health of each serv
48. se the application data in the VCS cluster is kept in a filesystem or volume manager not supported by SG LX you must migrate the application data to a filesystem or volume manager supported by SG LX Following are the steps to migrate a sample Apache service group from a VCS cluster to an SG LX cluster Plan SG LX heartbeat network and membership arbitration Collect the resources information used by your VCS service group Back up the application and confiquration data Install Servicequard and Apache toolkit software and prepare the nodes Create shared storage for application data Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Configure the SG LX cluster Stop the VCS cluster Start the SG LX cluster Verify the SG LX cluster Create and confiqure the Servicequard Apache toolkit package Start the Servicequard Apache toolkit package Verify the Apache toolkit package Remove the VCS configuration and software Plan SG LX heartbeat network and membership arbitration For SG LX cluster it is recommended to configure at least two networks or one channel bonded network for heartbeat In a migration you can also configure the network used by your VCS cluster for GAB LLT as heartbeat network for the target SG LX cluster VCS heartbeat network configuration is kept in the etc llttab file of each node A typical etc llttab file looks like below set node node 1
49. set cluster 2 link enl dev dlpi en 1 ether link en2 dev dlpi en 2 ether The third field of the lines starting with the directive link represents the network device used for GAB LLT by VCS SG LX heartbeat is at network layer TCP opposed to VCS heartbeat at link layer So you might need to configure an IP subnet or use an existing IP subnet on these network devices for SG LX heartbeat In this illustration the VCS cluster uses a channel bonded interface for GAB LLT The etc llttab looks like this set node hnode set cluster 26020 link bond0 bondO ether The target SG LX cluster can use an IP network over the bondO network interface as heartbeat network If the bondo interface is not configured with any IP network in the cluster nodes configure an IP network so that SG LX can detect and use the same for heartbeat For membership arbitration in the target SG LX cluster you can choose to use a cluster Lock LUN or Quorum Server When using Quorum Server for membership arbitration ensure that all the nodes can reach the server on the network f you choose to use a cluster Lock LUN ensure that you use a new disk Do not use any of the VCS coordination point disks or any other disk in use with the existing VCS cluster as a cluster Lock LUN in SG LX You can identify the coordination point disk being used by the VCS in the etc vxfentab file n this illustration the etc vxfentab can have the following
50. ster a er Server to HP Serviceguard for Linux he command can setup omatically specify the name of nthis i lustration the cluster na nerated cluster ascii file C cluster ascii GUI Serviceguard Manager Youcan also configure an SG LX cluster using Serviceguard Manager Please refer the section Configure the Cluster of HP SG LX Deployment Guide http bizsupport1 austin hp com bc docs support SupportManual c03393068 c03393068 pdf and follow the steps mentioned there to configure an SG LX cluster using Serviceguard Manager You can refer to cluster attributes comparison table VCS cluster attributes and SG LX cluster attributes and section Plan SG LX heartbeat network and membership arbitration to modify other SG LX cluster attributes to match you VCS cluster behavior Stop the VCS cluster Log in as root user and execute the following commands ttributes comparison tab Plan SG LX heartbeat network and membership arbitration to m he cluster Loc he cluster as t can automatically detect the network topology between the nodes and configure the eligible LUN using the disk passed through L option When using use the q Quorum Server name option instead e VCS cluster attributes and SG LX cluster attributes and section odify other SG LX cluster attributes to match you VCS he name of the node on which cmquerycl is run with a me is specified as apache cluster by modifying the clus
51. ter name owing command to apply the cluster ASCII file as root user Bring down the Apache agent service group on the hnode assuming it was running on the hnode hagrp offline apache agent servicegroup sys hnode e Check the status of Apache agent service group It should be offline hagrp state apache agent servicegroup Stop VCS on both nodes hnode and pnode and execute the following command on the both nodes hastop local Verify that the VCS engine port h is closed gabconfig a etc init d vxfen stop etc init d gab stop Jetc init d llt stop Start the SG LX cluster CLI Execute the following command to start the cluster cmruncl Stop 1 0 fencing if you have configured it Stop GAB and then LLT using following commands cmruncl Validating network configuration cmruncl Network validation complete Waiting for cluster to form done Cluster successfully formed Check the syslog files on all nodes in the cluster to verify that no warnings occurred during startup Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux GUI Serviceguard Manager Launch Serviceguard Manager from one of the node and go to the Administration menu and select Run Cluster Verify the SG LX cluster CLI Execute the following command to view the status of the cluster The status should be up
52. terlInterval are not relevant in SG LX Environment Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Mapping of VCS system attributes to SG LX node attributes The following table maps the VCS system attributes to the closest SG LX node attributes Table 3 VCS system attributes and SG LX node attributes VCS system attributes Corresponding SG LX node attributes Descriptions which can be editable by equivalent to some extent which can be users editable by users 1 Capacity capacity name and capacity value n VCS the Capacity attribute is configured to express total system load capacity This value is relative to other systems in the cluster and does not reflect any real value associated with a particular system n Serviceguard capacity name and capacity value attributes are used to define capacities of nodes 2 DynamicLoad Generic Resource framework n VCS the DynamicLoad attribute is used to indicate system maintained value of current dynamic load The value is set external to VCS with the hasys load command n Serviceguard Generic Resource framework is used for modeling dynamic metrics to influence failover 3 SourceFile cmclconfig file n VCS the SourceFile attribute indicates a file from which the configuration is read n Serviceguard configuration file is stored in binary format with he name cmclconfig 4 SystemLocatio
53. tero tiec t aas 27 Verify the Apache toolkit Package c ccccsssesssssssssssssssssssesssssecsscsesscasecnssessessssnccncaccscausssssseasancasascauecseassssssseasaneaseaneaeasenss 27 Remove the VCS configuration and software eene iisipin iaia 27 TEOUbleshOOLiig tips tto ette bec acc cuta dte dn na a t rea 28 Mr 28 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Executive summary This white paper describes a way to migrate Symantec Veritas Cluster Server VCS setup to HP Serviceguard for Linux SG LX setup It provides an overview of functionalities and features of VCS and SG LX and maps VCS paradigm to SG LX paradigm It also includes a step by step process to migrate an existing VCS 6 0 setup to a SG LX A 11 20 00 setup Introduction This paper helps users to migrate from a Veritas Cluster setup to a SG LX setup HP Serviceguard has been one of the preferred HA solutions of the industry for over 20 years and is available on HP UX Integrity and Linux x86_64 bit platforms HP Serviceguard solutions for Linux are built to address the financial business and operational impacts of downtime in a mission critical x86 infrastructure You can protect your mission critical workloads running on x86_64 bit platforms from downtime using SG LX It is a reliable cost effective proven high availability HA solution that is easy to deploy and manage It i
54. tion Guide for Linux 27 Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux Troubleshooting tips Refer to Troubleshooting Your Cluster of Managing HP SG LX http bizsupport2 austin hp com bc docs support SupportManual c03393076 c03393076 pdf Terms FC Fiber channel GAB Group Membership Services and Atomic Broadcast GUI Graphical user interface HA High availability HAD High availability daemon LVM Logical volume management facilities on Linux LLT Low Latency Transport os Operating system SG LX Serviceguard for Linux SPOF Single point of failure vcs Veritas Cluster Server For more information HP SG LX product documentation at hp com go linux servicequard docs HP SG LX certification matrix showing servers storage and software versions supported at hp com info sglx Veritas documents at https sort symantec com documents doc_details vcs 6 0 Linux ProductGuides Download SG LX evaluation software for FREE from following link https h20392 www2 hp com portal swdepot displayProductInfo do productNumber SGLX1120 DEMO The Serviceguard for Linux evaluation software can be used only for 90 days Learn more at hp com go sglx Sign up for updates hp com go getupdated Rate this document Copyright 2013 Hewlett Packard Development Company L P The information contained herein is subject to change without notice The only w
55. ty No attribute n VCS you can define the priority in which high availability daemon HAD runs by using the EnginePriority attribute n Serviceguard core daemons usually run at highest priority Technical white paper Migrating Symantec Veritas Cluster Server to HP Serviceguard for Linux VCS cluster attributes Corresponding SG LX Descriptions which can be configured cluster attributes by users equivalent to some extent which can be configured by users 6 Guests user_name user_host and In VCS the Guests attribute is used to specify a list of OS user accounts that have user_role Guest privileges on the cluster Serviceguard provides RBAC to specify Guest privilege to multiple OS users To achieve this USER ROLE is specified as MONITOR 7 GroupLimit max configured packages n VCS the GroupLimit attribute is used to specify maximum number of Service groups n Serviceguard the max configured packages attribute is used to specify the maximum number of packages 8 LockMemory No attribute n VCS the LockMemory attribute is used to control the locking of VCS engine pages in memory Serviceguard core daemons are pinned in the memory by default 9 LogSize No attribute n VCS the LogSize indicates the size of engine log files in bytes 10 OperatorGroups user name user host and n VCS the OperatorGroups and Operator attributes allow you to specify list of and
56. use Continentalclusters with other non HP disk array based replication In such cases Continentalclusters can be used to automate the recovery process and the user is responsible for data currency and consistency before initiating a recovery SG LX cluster membership An SG LX cluster is a networked grouping of HP ProLiant systems host systems known as nodes having sufficient redundancy of software and hardware that a single failure may not significantly disrupt service Application services individual Linux processes and their corresponding storage and network resources are deployed together using the package construct In the event of a single service node network or other resource failure SG LX can automatically transfer control of the package to another node within the cluster allowing services to remain available with minimal interruption Cluster quorum In the case of a failure SG LX cluster reforms with the surviving nodes In SG LX each cluster node h cluster reformation the nodes that can communicate with each other regroup and attempt a cluster can be reformed only if the cluster quorum condition is achieved The cluster quorum cond a strict majority that is more than 50 of the nodes running previously is available to form remaining nodes that are not in the group with cluster quorum condition can be reset Since y based on the previously running cluster membership and not o r SG LX clusters can reform even after rep
Download Pdf Manuals
Related Search