Network Troubleshooting
Contents
1. Troubleshooting by Othmar Kyas FDDI An Agilent Technologies Publication Lt ee Agilent Technologies st g FDDI Any time you think things seem to be going better you have overlooked something ANONYMOUS 9 1 FDDI Specification and Implementation Like Token Ring the Fiber Distributed Data Interface FDDI is based on a token passing principle With this technique access to the LAN medium is controlled primarily by means of a specific sequence of bits called a token Unlike Token Ring however FDDI uses a dual ring architecture for increased Station 1 MAC Ring Wrapping Ring Wrapping Station 4 Station 2 O A Bl O ca AE m lt B AEZ failed Station 5 Station 3 Figure 9 1 Fault tolerance in FDDI Ring recovery through wrapping TROUBLESHOOTING LocaL AREA NETWORKS 9 264 FDDI reliability and its greater bandwidth allows a data speed of 100 Mbit s The dual ring architecture enables FDDI to tolerate the complete failure of one of its nodes with no significant effect on network performance This fault tolerant feature is called ring wrapping if a station fails the ring doubles back on itself on either side of the failed station thus forming a single ring isolating the source of error and providing continuous data transmission see Figure 9 1 Two types of nodes are defined in2. Checks all internal data paths in the node e Performs loopback tests e Checks the parameters passed to the FDDI layer TTRT etc e Tests the FDDI recovery processes beacon claim token etc Physical Connection Management PCM PCM controls station output and the redundant fiber optic line to the neighbor ing node Together with the PCM of the neighboring node it tests the connection between the two nodes and checks the BER to determine whether a connection can be set up or not Each of the station s ports has its own PCM The 10 bits used in this bit signaling between nodes are explained in the following TROUBLESHOOTING LocaL AREA NETWORKS 9 280 FDDI Bit 0 Always set to 0 reserved for future applications Bits 1 2 Indicates the station s own interface type 00 A 01 B 10 S 11 M Bit 3 Compatibility of the output ports If Bit 3 0 for both ports the connection is not set up Bits 4 5 Link Confidence Test LCT This function tests the reliability of the connection How long the test takes depends on the previous BER 00 short 50 ms 01 medium 500 ms 10 long 5 s 11 extended 50 s Bit 6 Indicates whether the MAC layer is used during the LCT Bit 7 Indicates whether the BER was low enough to pass the LCT Bit 8 Indicates whether a MAC loopback test should be performed Bit 9 Indicates whether there is a MAC layer at the station output port Once the PCM has reached the active state the st
3. counters 1w3d Output queue 0 40 0 drops input queue 0 75 132 drops Five minute input rate 264000 bits sec 81 frames sec Five minute output rate 267000 bits sec 88 frames sec 33457636 frames input 2146812161 bytes 8 no buffer Received 2456722 broadcasts 0 runts O giants 15256 input errors 11561 CRC 176 frame O overrun 53676 ignored O abort 124789478 frames output 4146709113 bytes 379 underruns O output errors O collisions O interface resets O restarts 5460 transitions O traces 2405 claims 4 beacon If the problem cannot be isolated using the information described previously additional trend measurements are necessary This involves recording the main operating parameters over a period of hours or even days and analyzing the results for correlations In this way possible causes can be systematically eliminated until the source of the problem is limited to a small area The steps to take after the basic measurements have been performed using a protocol analyzer depend on the nature of the symptoms If the symptoms can TROUBLESHOOTING LocaL AREA NETWORKS 287 FDDI EY Internet Advisor LAN Capture Buffer FDDI Vitals Expert File Run View GoTo Setup Window Help 5 x al T TL Ls 4 pet tie E ee Jer Be Pre filter FDDI Utilization 46 0 Pre filter Frames oe p Te 1878 Z 0 Pre filter Tokens 0 0 0 0 900000 0 Pre filter LLC Frames 466 42 466 1878 20000 0 P
4. 05 00101 not be transmitted 06 00110 If received codes 1 2 8 and 16 08 01000 should always be interpreted as Halt 12 01100 16 10000 12345 Transmission order of code symbol bits Figure 9 3 4B 5B symbol encoding in FDDI TROUBLESHOOTING LocaL AREA NETWORKS 9 267 FDDI transmission ends with two ending delimiters If there is an even number of control indicators an additional ED follows the last control indicator Valid data symbols are the 16 hexadecimal values from 0 to F transmitted in any order Invalid symbols are any symbols that do not fulfill this definition A node may receive invalid symbols due to an error situation or during synchronization with the ring clock rate see Figure 9 3 9 1 1 1 FDDI Line States The line state is the fundamental indication of the operational status of the FDDI ring It is monitored continuously by each node s station management SMT entity The various line states are signaled as described here Quiet Line State QLS When a physical connection is first set up a steady stream of quiet Q symbols is transmitted QLS is also entered any time the signal is lost or after 16 or 17 consecutive Q symbols are received QLS ends when any symbol other than a Q is received Master Line State MLS MLS is indicated by a continuous stream of alternating halt H and quiet Q symbols and is also used to set up a new physical connection This state is entered whenever eight or nine consecu
5. frame because it ends with idle symbols TROUBLESHOOTING LocaL AREA NETWORKS 9 277 FDDI 9 1 3 Process Control in FDDI Networks Access to the medium in FDDI networks is controlled by a token timers and counters control various related processes A station with data to send must wait for a free token When the token is received transmission can begin Unlike Token Ring in which a transmitting node waits for its frame to return before releasing the token FDDI nodes release the token as soon as transmission is completed Each node is responsible for removing the frames that it transmitted from the ring When a node detects a frame containing its own address as the source it replaces the data field of the frame with idle symbols This results in frame fragments consisting of the PA SD FC DA and SA fields followed by idle symbols These remnants do not negatively affect ring performance however because they have recognizable defects such as the lack of an ending delimiter they are deleted by the next station that detects them while in the transmitting state Stations that are not in the transmitting state simply repeat and amplify the incoming bit stream A FDDI ring supports two types of communication synchronous in which each node is granted a defined portion of the available bandwidth and asynchronous with dynamic bandwidth sharing If a token is received by a node before the node s TRT has reached the TTRT this is known
6. FDDI dual attachment stations DAS and single attachment stations SAS An SAS is attached to the primary ring through a concentrator similar to the concentrators used in Token Ring An SAS re quires only a single FDDI port and can be inserted into or removed from the ring without affecting network operation Dual attachment stations require two ports attached to both the primary and secondary rings Connecting or disconnecting a DAS disrupts ring operation FDDI FDDI connection options i Dual attached stations DAS FA Ta f and single attached stations SAS SAS SAS SAS Primary Primary Port B Secondary Secondary Connector assignments FDDI DAS on the DAS interface Figure 9 2 Station connections in FDDI Dual homing is another fault tolerant feature of FDDI for connection of critical devices such as servers and routers With dual homing the component is attached to two concentrators The second concentrator link remains passive unless the primary link fails FDDI protocols can be implemented over two pairs of single mode or multimode optical fibers or over four pairs of shielded or unshielded twisted pair copper TROUBLESHOOTING LocaL AREA NETWORKS 9 265 FDDI wires STP UTP FDDI over copper wire is called Copper Distributed Data Interface CDDI 9 1 1 The Physical Layer in FDDI Networks The FDDI physical layer is defined for single and multimode optica
7. Frequent ring initializations and high bit error rates are often symptoms that the signal power of a NIC or concentrator is too weak To determine whether this is the case measure the power at a node s receiving port when a constant stream of Halt symbols is transmitted The average must be at least 20 dBm Cause 1 Loose connectors dust or fingerprints on optical fiber or connector Cause 2 DAS deactivated If a dual attachment station or concentrator fails or is deacti vated the distance between two stations may exceed the maxi mum specifications In a network with high redundancy the ring should be designed so that no two neighboring nodes are more than 400 meters apart Then the ring can remain operational even if up to four contiguous stations fail Cause 3 Active optical bypass switch Optical bypass switches are activated when a node fails and can increase attenuation caused by the ring by up to 2 dB If several bypasses are active the resulting loss can lead to high bit error rates and consequent increases in claim and beacon frames 294 Cause 4 Cause 5 Symptom Cause 1 Cause 2 Symptom Cause Symptom Cause Symptom Cause Symptom Cause TROUBLESHOOTING LocaL AREA NETWORKS 9 FDDI Defective interface card Defective port in a router bridge or concentrator Large Number of Status Report Frames New MAC neighbor Change in port s operating status FDDI sta
8. control indicators are optional E A C See ee ae R S RS RS RS RIS T la kl Is Terminate symbol R S Status bit set 1 or reset 0 Figure 9 9 The frame status field in FDDI Error Detected E If a ring node detects an error in the frame it sets the value in the E field to S Address Recognized A When a node recognizes the destination address of a frame as its own it sets the value in the A field to S Frame Copied C When the receiving node copies the frame into its receive buffer it sets the value in the C field to S TROUBLESHOOTING LocaL AREA NETWORKS 9 275 FDDI 9 1 2 2 Timers and Counters Every FDDI node has three timers for controlling ring activities The values set for these timers are calculated using the following parameters D_Max maximum ring delay default 1 617 ms M_Max 1 000 maximum number of FDDI interfaces in the ring I_Max 25 0 ms maximum station insertion time A_ Max 1 0 ms maximum signal access time Token_Time 0 00088 ms the time it takes to transmit a token 6 symbols with preamble 16 symbols L_Max 0 0035 ms maximum time between receipt of token and start of transmission F_Max 0 361 ms maximum time for transmitting a data packet equals the transmission time for 9 000 symbols plus 16 preamble sym bols Claim_FR 0 00256 ms time required for transmission of a claim token frame S_Min 0 3545 ms the time it takes for the ring to
9. detail in the chapter on Ethernet networks the following discussion deals only with bridge problems specific to FDDI 9 2 6 1 Diagnosing Bridge Problems The challenge in analyzing bridge problems is to correlate the occurrence of symptoms in several different network segments It is not as important to measure network performance in several LAN segments simultaneously Perfor mance measurements would only be required to determine throughput or trans mission delay It is more efficient to request system specifications from the manufacturer based on standardized test methods as specified in RFC 1242 and RFC 1944 TROUBLESHOOTING LocaL AREA NETWORKS 9 292 FDDI Most problems that affect bridges can best be located by a process of elimination that involves the correlation of specific measurements and an analysis of the network topology Symptoms of bridge problems can include poor network performance in particular segments intermittent or permanent loss of connec tions to particular nodes or the failure of certain protocols and services The first phase of the troubleshooting process is as always a review of all configura tion changes that were made in the network before the error occurred as well as the general information gathering steps described previously If the symptoms correlate to particular connections begin by checking all bridges located along the corresponding transmission path Otherwise the next step is to prepare a lis
10. of Os is called a zero address frames with this address are not intended for any station The second bit in a destination address shows whether the address is locally administered bit 2 1 or universally adminis tered bit 2 0 Figure 9 8 shows a FDDI data packet that has been decoded using a protocol analyzer Occasionally FDDI addresses are shown in both the MSB Most Significant Bit First format and the canonical format used in Ethernet and other network protocols To convert between the canonical bit order and MSB format the nibbles half bytes are switched and the bit order in each half reversed Thus a hexadecimal 43 becomes 34 or in binary 0011 0100 Reversing the bit order of each nibble yields 1100 0010 or in hexadecimal C2 The canonical address 01 80 C2 00 01 10 corresponds to an MSB address of 80 0 1 43 00 80 08 Internet Advisor LAN File Decode Decode Data demo2fdi dat M Eile Run View GoTo Setup Window Help 5 x a B N 10 04 56 92877033 raj em Rect Time C M Summary M Detailed F Hex ASCII C EBCDIC Filter S UL al Proto IP UL 5 Proto IP FDDI Header FDDI FC OxDO FDDI 1 Class Of Service Synchronous a FDDI 1 Address Length 48 bits FDDI 01 0000 Frame Type LLC Frame FDDI Destination Address 3Com 86800C 40063186800C FDDI 0 Individual Address FDDI 1 Locally Administered Address FD
11. s T_Min value is higher than T_Opr the station cannot participate in normal ring traffic T_Min 4 0 ms default T_Max 4 T_Init gt 165 ms where T_lnit is the time the ring has operated without noise T_Init T_React T_Resp lt 40 58 ms where T_React lt Max D_Max A_Max TVX and T_React lt 30 24 ms T_Resp lt 3 D_Max 2 M_Max Claim_FR S_Min and T_Resp lt 10 34 ms Late_Ct is set to 1 when the node is initialized or reset and is incremented every time the TRT runs out without a token having been received Once a token is received the Late_Ct is reset to 0 To simplify troubleshooting and the isolation of failure domains in the ring every FDDI station has counters that count every data packet whether defective or not However the frames are counted only if they end in an ending delimiter T symbol Data packets that end with idle or invalid symbols are not counted Frame_Ct Counts all frames received Error_Ct The number of frames identified by this node as defective in other words those frames whose error detected E field is R on arrival at this node but S on retransmission Frames received with E already set to S are not counted Lost Ct The number of frames including tokens that are in the process of being received by a station when an error occurs The lost counter is incremented and the rest of the frame is replaced with idle symbols The next node does not count this
12. DI Source Address HP 90B6AC 10009090B6AC FDDI Size Of Info Field 560 FDDI FCS 07BAE2E2 pes i aria LLC Header LC Dsap Oxaa 170 LC Ssap Oxaa 170 Command LC Unnumbered frame UI 3 SNAP Header x Ready FDDI Off Line OM Fiber DAS AISO BISO Figure 9 8 Decoding an FDDI frame using the Agilent Advisor protocol analyzer TROUBLESHOOTING LocaL AREA NETWORKS 9 274 FDDI Information Field The information or data field contains user data the payload of the FDDI frame The data type is described in the frame control field and evaluated accordingly by the receiving node s MAC LLC or SMT module The length of the data field is variable but the length of the entire frame must not exceed 9 000 symbols or 4 500 bytes Frame Check Sequence FCS This is a 32 bit checksum calculated from the content of the frame control source address destination address and data fields Each receiving station evaluates the checksum Ending Delimiter ED The ending delimiter marks the end of a token or data packet This field consists of two consecutive T symbols in a token or one T in a data packet Frame Status FS The frame status field consists of control indicators that follow the ending delimiter The first three control indicators E error detected A address recognized and C frame copied are required and are set to R reset by the source station on transmission Other
13. SHOOTING LocaL AREA NETWORKS 9 296 FDDI Common Errors The following list summarizes the most frequent sources of problems in FDDI networks in alphabetical order e Bridge address list incorrectly configured bridge in protected mode e Bridge filter incorrectly configured e Bridge overloaded e Bridge s aging function deletes address entry e Cable length between neighboring nodes exceeds specifications especially after a DAS node failure or ring wrapping e CDDI only electromagnetic interference e Connectors loose or defective interface cards wall jacks concentrators bridges routers e Defective patch cable e Defective concentrator e Duplicate FDDI ring addresses e Faulty physical installation of router bridge or concentrator loose cable connectors plug in modules incorrect cable connections on the backplane e Fiber only dust or fingerprints on the connector e Frame length restrictions on router bridge ports e Frequency and jitter problems due to cabling noise too many stations e Network interface card defective e Network interface cards incorrectly configured TTRT driver interrupt e Receive buffer on interface card insufficient e Router filter incorrectly configured e Router overloaded e Router protocol entries incorrectly configured address tables mapping tables subnet masks default gateways routing tables timers e Router settings incorrectly configured port not active protocol not act
14. _Pri values As soon as a node captures a token its THT is initialized with the value remaining in the TRT The TRT itself is initialized with the current value of TTRT so that the time of the next token rotation is measured relative to the target time If a node has a large amount of data to transmit in a short time it initiates the restricted token state With this technique the node first captures a nonre stricted token and begins its data transmission When the THT runs out the node issues a restricted token which is simply forwarded by all other nodes until it is returned to its source node The restricted token state lasts until completion of the transmission for which it was started usually a period of several TRTs While the ring is in this state all other asynchronous transmission is stopped Synchronous transmission however which uses both types of token is not affected The maximum duration of the restricted token state is negotiated by the SMT Claim Token Process All nodes monitor the ring for errors that necessitate reinitialization of the ring such as inactivity when the TVX runs out or signal errors if TRT runs out and Late_Ct is already set for example When a node detects such an error it sets the ring operational variable to 0 and transmits a claim token frame indicating its desired TTRT It begins checking the TTRT values of all claim token frames it receives The lower the TTRT value the higher the sender s
15. a percentage of capacity throughput in frames per second token rotation time the numbers of stripped and void frames the numbers of claim and beacon frames the number of frames with undersized preambles and the number of frames with invalid checksums TROUBLESHOOTING LocaL AREA NETWORKS 9 286 FDDI The analysis of these statistics often points to possible causes of the problem Furthermore all SMT frames should be recorded and analyzed including SRF and RDF which can point to the failure domain Furthermore you can use the SHOW INTERFACES FDDI command exact command depends on the equip ment type to check the statistics of the interface cards of ring nodes The following is a sample result of a SHOW INTERFACES FDDI command entered on a DAS Fddi O is up line protocol is up Hardware is cBus Fddi address is 0000 0b14 32e2 bia 0000 0b14 32e2 Internet address is 18 187 1 29 subnet mask is 255 255 254 0 MTU 4470 bytes BW 100000 Kbit DLY 100 usec rely 255 255 load 1 255 Encapsulation SNAP loopback not set keepalive not set ARP type SNAP ARP Timeout 3 00 00 Phy A state is active neighbor is B cmt signal bits 008 20C status ILS Phy B state is active neighbor is A cmt signal bits 20C 008 status ILS CFM is thru A token rotation 5000 usec ring operational 2 13 46 Upstream neighbor 0000 7640 0e50 downstream neighbor 0000 0a02 5bf2 Last input 0 00 00 output 0 00 00 output hang never Last clearing of show interface
16. activated in the manufacturer s default configuration x connection not permitted 9 1 5 FDDI Standards ANSI X3 139 1987 ISO 9314 2 1989 Media Access Control MAC ANSI X3 148 1988 ISO 9314 1 1989 Physical Layer Protocol PHY ANSI X3 166 1990 ISO 9314 3 1990 Physical Layer Medium Dependent PMD ANSI X3 229 1994 ISO 9314 6 Station Management SMT ANSI X3 184 1993 ISO 9314 4 Single Mode Fiber PMD SMF PMD ANSI X3 237 1995 ISO 9314 9 Low Cost Fiber PMD LCF PMD ANSI X3 263 1995 ISO 9314 10 Twisted Pair PMD TP PMD ANSI X3 278 Physical Layer Repeater PHY REP ANSI X3 262 ISO 9314 13 Conformance Test PICS Proforma for FDDI CT PICS ANSI X3 245 199x ISO 9314 26 Abstract Test Suite for MAC MAC ATS ANSI X3 248 199x ISO 9314 21 Abstract Test Suite for PHY PHY ATS ANSI X3 255 199x ISO 9314 20 Abstract Test Suite for PMD PMD ATS ANSI X3T9 5 92 102 Rev 1 4 Abstract Test Suite for SMT SMT ATS RFC 1285 FDDI MIB TROUBLESHOOTING LocaL AREA NETWORKS 9 285 FDDI 9 2 Troubleshooting FDDI Networks 9 2 1 Gathering Information on Symptoms and Recent Changes The first step in any troubleshooting process is to gather information The more information you have about the symptoms and characteristics of a problem including when it first occurred the better your chances of solving the problem quickly and efficiently Typical questions you might ask at this stage include Do the symptoms occur regula
17. an answer Status Information Frame SIF SIFs provide information about the status of a node There are two types of SIF e SIF configuration frames describe the station s current configuration in cluding the number of input and output ports the number of interfaces and information on neighboring nodes e SIF operation frames describe the current operating state of a node includ ing MAC parameters LEM status of the ports and frame counters SIFs can be transmitted as request or response frames An SIF configuration response frame can contain up to 10 parameters including time stamp station descriptor SMT versions supported station state station policy added trans mission delay neighboring nodes path descriptors and parameter change count TROUBLESHOOTING LocaL AREA NETWORKS 9 282 FDDI SIF operations response frames can contain the following time stamp MAC status port LEM status MAC frame counter MAC frame not copied counter MAC priority values elasticity buffer status vendor code user field and param eter change count Echo Frame ECF When an ECF is received the SMT copies the data field and returns it as an echo response frame Request Denied Frame RDF When the SMT receives a data packet with an unknown format or with an SMT version it does not support it transmits an RDF Other causes for denial of a request may be oversized frames or the lack of reception authorization Status Report Fram
18. an take to travel around the entire ring Furthermore the station assumes the exist ence of a duplicate address if it receives a claim token frame with its own address as the source but a TTRT that differs from its own e When the MAC layer is active the RMT checks for duplicate addresses by verifying the A bit in neighbor information frames If it detects a duplicate address the RMT reacts in one of three ways it closes down the connection removes the station from the ring or changes the FDDI MAC address e The RMT also detects and responds to beacon states e It controls and initiates Halt Line States e It supports and monitors restricted tokens When a station receives a re stricted token the RMT starts a timer to monitor the duration of the re stricted access dialog If the timer expires a claim token or beacon process is triggered SMT Agents In addition to the four functional modules described previously every SMT also has an SMT agent that checks all incoming FDDI frames and acts on them if necessary The SMT agents use a number of special FDDI ring management frames in performing their tasks Neighbor Information Frame NIF NIFs are used to determine the identities of neighboring nodes Each node broadcasts a NIF approximately every 30 seconds The first station to receive an NIF with the A bit address recognized cleared is the nearest downstream neighbor of the node that sent that frame this neighbor transmits
19. and transmits beacon frames continuously A node that is not in the beacon state repeats any beacon frames it receives When a node receives its own beacon frames it assumes that the ring has recovered and begins the claim token process again 9 1 3 3 The FDDI Station Management Specification SMT SMT in FDDI is a special functional module integrated in the FDDI protocol stack that provides increased security and automatic fault recovery mecha nisms SMT includes the following functions e Inserting the node in the ring and removing it e Reconfiguring the paths in the node when a link fails for example e Checking the physical connection before inserting the node Controlling node behavior during the beacon process e Reporting the current node configuration e Transmitting status report frames to isolate possible error sources SMT is composed of four modules Entity Coordination Management ECM Physical Connection Management PCM Configuration Element Management CEM and Ring Management RMT Entity Coordination Management ECM ECM controls the optical bypass system as well as all other SMT functions As soon as an FDDI node becomes active ECM deactivates the optical bypass and starts all other SMT functions Similarly when the node leaves the ring ECM first stops all other SMT functions and then reactivates the optical bypass In the context of these processes ECM also performs a number of tests on the physical layer
20. arameter Management Frame PMF 283 Physical Connection Management PCM 279 Q Quiet Line State OLS 267 R Repeat filter 269 Request Denied Frame RDF 282 Resource Allocation Frame RAF 283 Ring delay 269 Ring Management RMT 280 S Single Attachment Station SAS 264 Smoothing 268 SMT agents 281 Status Information Frame SIF 281 Status Report Frame SRF 282 Symbol encoding in FDDI 266 Synchronous transmission 277 T Timer Valid Transmission TVX 275 Token Holding Timer THT 275 Token Rotation Timer TRT 275 TROUBLESHOOTING LocaL AREA NETWORKS 9 Index p3 FDDI V Violation frames 295 Void frames 290 W Wedged interface 295
21. as an early token the token can be used for synchronous or asynchronous transmission If the token is late however Late_Ct is increased by one the TRT is initialized with the value for T_Opr and the node may only transmit synchronously The Late_Ct is reset to 0 and asynchronous transmission is allowed only after a token has been received within the TTRT This ensures an average synchronous response time lt TTRT and a maximum synchronous response time of 2 TTRT 9 1 3 1 Synchronous Transmission In synchronous transmission every station is assigned a certain bandwidth expressed as a percentage of the TTRT This bandwidth allocation is 0 when a node is initialized a higher value is then negotiated by the SMT The sum of all allocated bandwidths must not exceed the maximum usable synchronous band width Bsyn_Max Bsyn_Max TTRT D_Max F_Max Token_Time 9 1 3 2 Asynchronous Transmission There are two types of tokens for asynchronous transmission nonrestricted tokens which are available to all ring nodes and restricted tokens which are reserved for certain nodes When the ring is re initialized a nonrestricted token is issued At this point priority levels can be distinguished by assigning TROUBLESHOOTING LocaL AREA NETWORKS 9 278 FDDI T_Pri values A node can capture a nonrestricted token only if the node s T_Pri is higher than the TRT Thus heavy ring traffic can be relieved to a certain extent by defining low T
22. ation begins transmitting either QLS signals or data The PCM also starts the link error monitor LEM which checks the BER in the FDDI port and deactivates the port if the BER is too high When the BER reaches 10 a warning is sent if it goes up to 107 the connection is shut down Configuration Element Management CEM CEM configures the station s internal data paths including the primary second ary and local paths For this purpose each port has a logical module called the configuration control element CCE which distributes incoming data among these internal paths When the CEM changes the status of a CCE it also deletes all data in the ring by transmitting ILS signals This causes the ring nodes to begin transmitting claim token frames Ring Management RMT RMT controls the FDDI protocol stack Itis not active until a physical connection exists and the input and output ports have been assigned to internal data paths The RMT has six main tasks e It initializes the MAC layer once a physical connection has been set up e When the MAC layer is not active it checks for duplicate addresses by monitoring claim token and beacon processes To do this the station s RMT TROUBLESHOOTING LocaL AREA NETWORKS 9 281 FDDI evaluates all claim token and beacon frames if a node receives its own claim token or beacon frame after more than 2 D_Max another station has the same address D_Max is the maximum length of time a frame c
23. dress length bits format bits and control bits Frame Class bit C 0 Frame Class bit C 1 Address Length bit L 0 Address Length bit L 1 The frame is asynchronous The frame is synchronous 16 bit MAC addresses 48 bit MAC addresses The frame format bits FF together with the C L and ZZZZ bits indicate the frame type as follows CLFF bits ZZZZ bits 0X00 0000 Void Frame content is ignored 1000 0000 Unlimited token 1100 0000 Limited token OLOO 0001 1111 Station Management frame 1L00 0001 1111 MAC frame 1L00 0010 MAC beacon frame 1L00 0011 MAC claim token frame CL01 r000 rlll LLC frame OLOI RPPP Asynchronous transmission with priority LLC 1L01 Rrrr Synchronous transmission LLC OLOO 0001 1111 SMT frame OLOO 1111 Next Station Addressing SMT frame CL10 r000 r111 Reserved for implementation CL11 Rrrr For future standardization X Any value r Reserved and set to 0 L Length C Class TROUBLESHOOTING LocaL AREA NETWORKS 9 272 FDDI The control bits in conjunction with the corresponding CLFF bits have the following meanings MAC Beacon Frames 1L00 0010 MAC beacon frames are transmitted when the ring is unable to recover from an error situation usually a hardware fault that results in signal failure jabber frames or frequency differences MAC Claim Token Frames 1L00 0011 These frames are usually transmitted when a token is lost When a station receives a claim token frame containing its own address as the sou
24. e SRF These frames report changes in the station s status including any of the follow ing events e Change in configuration Unwanted connection attempts e MAC neighbor change e MAC Frame error condition e MAC Path change condition e Port Path change event e Port Link error rate condition e Port Wrapping in a neighboring station e MAC Frame not copied e MAC Duplicate address e Port Elasticity buffer error e Vendor specific events The hold off and back off timers ensure that the station is not flooded with SRFs The hold off timer prevents transmission of status change reports more than once every 2 seconds while the back off timer controls the interval be tween change report transmissions Because no acknowledgement is sent in response to these frames the SMT repeats SRFs at ever increasing intervals 2 4 8 16 32 seconds TROUBLESHOOTING LocaL AREA NETWORKS 9 283 FDDI Parameter Management Frame PMF Network management devices can use these frames to read or change certain SMT variables The parameter management process corresponds to the struc ture of network management protocols such as SNMP or CMIP There are two types of PMF e PMF Get e PMF Set Extended Service Frame ESF The ESF format can be used to define custom SMT frames Resource Allocation Frame RAF RAFs are used to allocate synchronous bandwidth 9 1 4 Design Guidelines for FDDI Networks The guidelines for design
25. e combination of TTRT and claim counter values can be used to trace the node that initiated the claim process 289 _ Station C L TTRT 5 Station A Station B TTRT 10 TTRT 15 Claim Initiator Claim Counter A Claim Counter B Claim Counter C E A 1 oo 0 a 1 B 0 1 1 C 0 0 1 Figure 9 11 Identifying the node that initiated a claim process in an FDDI network TROUBLESHOOTING LocaL AREA NETWORKS 9 290 FDDI 9 2 3 2 Void Frames A void frame is one with the value 0X00 0000 in its frame control field where X stands for the address length bit and has a value of either 1 indicating a 48 bit address or 0 indicating a 16 bit address Void frames are not actually data packets and are usually ignored by all ring nodes Some manufacturers use void frames for special purposes however such as deleting frame fragments or stripped frames Thus the occurrence of void frames does not necessarily indi cate an error condition Contact the manufacturer of the components in ques tion for further details 9 2 4 Cabling Problems As in other networks cabling problems are frequent causes of errors in FDDI networks Typical causes include defective or low quality cables cable lengths exceeding the specified maximum defective or low quality connectors or in a CDDI network incorrect impedance or electromagnetic interference noise caused by air conditioning systems photocopiers pagers ele
26. frequent reinitialization of the ring ring state transitions and large numbers of lost token errors accompanied by increasing numbers of claim and beacon frames A physical break in the cabling or a power outage in a station or concentrator usually triggers ring wrapping You can usually identify the node at which the wrap occurred by analyzing status information frames with a protocol analyzer or by querying the status of interfaces such as bridge or router ports Then you can test all the components connector cable concentrator interface cards bridge router ports that were cut out of the ring when it wrapped Large numbers of claim and beacon frames in the ring in conjunction with frequent transitions may indicate problems either in the cabling kinks con taminated connectors or in the transmit and receive ports of an interface card In such cases trend measurements of the relevant parameters concurrent tracking of active stations and error rate and an analysis of status information frames can be useful in tracing the fault 9 2 3 1 Principal Error Conditions During Normal Ring Operation Claim Initiator Identification State transitions in an FDDI network do not necessarily constitute a sign of trouble in the ring Ring reinitialization is usually triggered by a node s LEM function when the error rate in its interface exceeds a certain threshold In this case the ring is temporarily deactivated while the LEM tests the link in
27. hing function can increase the length of a 0 to 13 symbol preamble to 14 symbols and reduce a preamble of 15 symbols or more to a length of 14 symbols Frames with preambles shorter than 12 symbols are TROUBLESHOOTING LocaL AREA NETWORKS 9 269 FDDI usually not forwarded on the FDDI layer and frames with a preamble of less than 2 symbols are ignored altogether 9 1 1 4 Repeat Filter If a station in the ring is acting as a repeater but the FDDI protocol stack does not check incoming signals the repeat filter prevents the propagation of code violations and invalid line states 9 1 1 5 Ring Delay To ensure trouble free operation of the ring every station with an FDDI MAC layer must have a minimum delay of 3 bytes while stations without an FDDI MAC layer must guarantee a delay of 2 bytes The resulting maximum ring delay is the sum of the delay caused by all stations and the signal delay inherent in the medium Both the MAC layer and the SMT have timers that take this figure for maximum ring delay into account The following parameters are used in calcu lating the overall delay SD_Min The minimum latency of a starting delimiter sequence in a station default 74 bits or 592 ns SD_Max The maximum latency of a starting delimiter sequence in a station The maximum extension due to the smoothing function is 2 symbols or 10 bits the elasticity buffer may add a similar delay 4 5 bits maximum addition of 9 bits Sampling and
28. idge Port Configured with Duplicate FDDI Address Because FDDI addresses are configured by software the occurrence of dupli cate addresses due to incorrect configuration typing errors copied configura tion files is not uncommon Inefficient Maximum Frame Length Incorrectly configured bridge ports that restrict the maximum frame size can have a negative effect on performance TROUBLESHOOTING LocaL AREA NETWORKS 9 293 FDDI Installation and Configuration Errors Among the leading causes of problems with bridges are incorrect installation or configuration of the equipment especially in the use of increasingly complex modular bridges Incorrectly configured ports FDDI interface not activated bad connections loose cables connectors or plug in modules and faulty con nections to the back plane or the wiring cabinet are the most common error sources Hardware Problems If you suspect hardware problems check the power supply and connectors and run the bridge s self test function 9 2 7 Problems with Routers Routers are internetworking components that connect network segments on OSI Layer 3 Because they operate on this layer routers can link networks of any topology Refer to the section on router problems in Chapter 7 for a detailed description of procedures for troubleshooting and diagnosing router errors 9 2 8 Symptoms and Causes FDDI Symptom Frequent Ring Reinitialization High Bit Error Rate Detected by LEM
29. ing FDDI networks include specifications for the vari ous cable types as well as limits on the maximum distances between neighboring nodes and the maximum number of nodes per ring The maximum distance between two adjacent nodes is 2 km on multimode fiber rings 40 km on single mode fiber rings and 500 meters in low cost fiber LCF rings Itis important to keep in mind that when a ring wraps due to node failure the ring length doubles The wavelength used in all fiber optic rings is 1 300 nm The specifications for diameter and signal power are as follows Multimode Diameter 62 5 125 mm 50 125 mm 85 125 mm 100 140 mm Signal power 14 dBm to 20 dBm Single mode Diameter 9 125 mm Signal power 14 dBm to 20 dBm Category 1 15 dBm to 37 dBm Category 2 When shielded STP 1 or unshielded UTP 5 twisted pair cabling is used the maximum distance between two nodes is 100 meters There are no values defined for minimum distances between nodes in either FDDI or CDDI 9 1 4 1 Connection Rules for SAS and DAS Nodes When connecting a dual attachment station DAS port A of one DAS must be connected to port B of the neighboring node For single attachment stations TROUBLESHOOTING LocaL AREA NETWORKS 9 284 FDDI SAS the S port of the node must be connected to the M port of the concentra tor A B T M X S gt recommended connection connection could lead to problems may be de
30. insertion or removal process from beginning to end completion is defined as 1 5 dB below the final signal level Other problems in the physical layer that are not specific to FDDI and their remedies are discussed in detail in Chapter 5 9 2 5 Problems with FDDI Interface Cards Typical symptoms of defective interface cards in FDDI rings are high numbers of claim and beacon frames in conjunction with frequent transitions The first step in localizing a defective FDDI NIC is to identify suspicious nodes on the network Begin by making a list of all network nodes that transmit defective frames Most protocol analyzers provide this information with fully automatic test programs You can also use the method described in Section 9 2 3 to determine which node initiates the claim process If this does not pinpoint the problem or if the symptoms are intermittent try the correlation method begin by simultaneously charting the activity of the suspicious nodes and the error rate in the network If there is a correlation between the activity of a certain node and the error rate then you have probably found the defective interface card 9 2 6 Problems with Bridges Bridges are components that connect network segments on OSI Layer 2 the MAC layer Bridges buffer and filter the frames they receive from connected segments and transmit them to their destination segments without regard to higher layer protocols The basic functions of bridges are described in
31. ive e Signal loss due to active optical bypass switch e Stations too many on the ring e WAN connections down overloaded or of poor quality high BER Figure 9 12 The most common causes of errors in FDDI networks TROUBLESHOOTING LocaL AREA NETWORKS 9 Index p1 FDDI Index of chapter 9 A Active Line State ALS 267 Asynchronous transmission 277 B Beacon process 278 C Claim initiator identification 288 Claim token process 278 Configuration Element Management CEM 280 Connection rules for SAS and DAS nodes 283 D Design guidelines for FDDI networks 283 Dual homing 264 Dual Attachment Stations DAS 264 Duplicate FDDI address 292 E Echo Frame ECF 282 Elasticity buffer 268 Entity Coordination Management ECM 279 Error symptoms in FDDI 288 Extended Service Frame ESF 283 F Fault tolerance in FDDI 263 FDDI 263 FDDI data format 269 FDDI frame 270 FDDI frames with the error bit set 294 FDDI interface cards 291 FDDI MAC layer 269 FDDI protocol 265 FDDI Station Management Specification SMT 279 FDDI token 270 Fiber Distributed Data Interface FDDI 263 Frequent ring reinitialization 293 TROUBLESHOOTING LocaL AREA NETWORKS 9 Index p2 FDDI H Halt Line State HLS 267 I Idle Line State ILS 267 M Master Line State MLS 267 Multi Level Transition Three Level Technique MLT 3 265 N Neighbor Information Frame NIF 281 Noise Line State NLS 268 P P
32. l fiber as well as for shielded and unshielded twisted pair copper wire CDDI Transmis sion is limited to defined symbols in 4B 5B encoding When an optical data medium is used 4B 5B encoded data streams are transmitted directly in the form of light pulses With twisted pair wiring the Multi Level Transition Three Level Technique MLT 3 is used MLT 3 alternates between three voltage levels reducing the frequency of the transmitted signal to 31 25 MHz FDDI and CDDI also have different idle signals in FDDI a bit stream consisting of binary 1s indicates an idle line station whereas in CDDI the idle signal is a random series of 1s and Os because a continuous sequence of 1s would distort the frequency spectrum and increase electromagnetic interference The FDDI protocol uses three types of symbols e Line state symbols indicating one of the following e Quiet Line State QLS e Master Line State MLS e Halt Line State HLS e Idle Line State ILS e Active Line State ALS e Noise Line State NLS e Control symbols including the starting delimiter ending delimiter and control indicators e Data symbols Line state symbols are sent as padding bits during pauses in transmission and indicate the operating state of the FDDI ring Halt symbols for example either announce control sequences or report the removal of invalid symbols while at the same time minimizing any DC imbalance in signals on a CDDI ring Quiet symbols report an abse
33. nce of voltage transition which means there is no signal in the line Idle symbols indicate a normal operating state between transmis sions These consist of continuous padding bits which provide clock informa tion for synchronization The starting delimiter SD and ending delimiter ED control symbols mark the beginning and end of a transmitted data sequence The ending delimiter how ever is not necessarily the last symbol in a transmission it may be followed by a set S or reset R control indicator If no control indicators are sent then the TROUBLESHOOTING LocaL AREA NETWORKS 9 266 FDDI FDDI Symbol Coding Decimal Binary Symbol Description Line state symbols 00 00000 Quiet 31 11111 Idle 04 00100 Halt Starting delimiter 24 11000 First symbol of the SC pair 17 10001 Second symbol of the SD pair Data symbols Hexadecimal Binary 30 11110 0 0 0000 09 01001 1 1 0001 20 10100 2 2 0010 21 10101 3 3 0011 10 01010 4 4 0100 11 01011 5 5 0101 14 01110 6 6 0110 15 01111 T 7 0111 18 10010 8 8 1000 19 10011 9 9 1001 22 10110 A A 1010 23 10111 B B 1011 26 11010 C C 1100 27 11011 D D 1101 28 11100 E E 1110 29 11101 F F 1111 Ending delimiter 13 01101 Marks the end of the data stream Control markers 07 001111 Logical 0 reset 25 11001 Logical 1 set Invalid codes 01 00001 These symbols violate the conditions 02 00010 for zero bits in the code stream or 03 00011 the mandatory sequence and should
34. priority If the TTRTs of two frames are equal the one with the higher source value has higher priority When a node detects claim token frames with a higher priority level than its own it stops issuing claim token frames Eventually the ring contains only claim token frames from the node with the lowest TTRT This node initializes the ring resets T_Opr to its own TTRT starts the TRT and issues a nonrestricted token If a station s TRT expires before another higher priority node initializes the ring then this station begins sending claim token frames again rather than sending beacon frames This prevents sporadic beacon frames in the ring The token cannot be captured by any station during its first rotation because the ring operational variable is cleared when the claim token process starts Once the first rotation has been completed both Ring_Operational and Late_Ct are set to 1 and TRT is initialized Synchronous transmission can begin in the second token rotation asynchronous transmission in the third Beacon Process If a node s TRT runs out while a node is in the claim token state the node considers the claim token process to have failed and starts the beacon process TROUBLESHOOTING LocaL AREA NETWORKS 9 279 FDDI As a rule this only happens when the ring is physically interrupted and must be globally reconfigured when one logical ring is broken into two for example After entering the beacon state the node resets its TRT
35. ques tion The ring is also reinitialized any time a node s TVX expires indicating that TROUBLESHOOTING LocaL AREA NETWORKS FDDI 9 the node has not received a token or other valid data packet in the last 2 5 ms The reinitialization procedure takes only a few milliseconds so the higher layer protocols with timer values on the order of whole seconds are not affected If an error that triggers reinitialization occurs as part of normal operation when a node is connected or disconnected a router or bridge is rebooted etc then the reinitialization process does not indicate a problem If transitions happen sev eral times a minute however then the higher layer protocols are affected If you can localize the station that started the claim process you should be able to isolate the failure domain and solve the problem To determine which station initiated a claim process assign a different TTRT value to each station in the ring To illustrate this method consider a ring with three nodes Node A has TTRT 10 Node B has TTRT 15 and Node C has TTRT 5 Analyze the claim counters Claim_Ct in each station In this example assume Node B starts the claim process Node A repeats the frame transmitted by Node B because TTRT gt TTRT Node C however replaces B s frame with its own claim frame because TTRT lt TTRT The claim counters of the three stations contain the following values at this point B 1 C 1 A 0 In this way th
36. rce it reinitializes the ring and issues a new token SMT Next Station Addressing Frame OLOO 1111 This frame is used for station management functions LLC Frame OLO1 rPPP This LLC frame is used for asynchronous transmission The last three bits PPP indicate the priority The highest priority is 111 and 000 is the lowest LLC Frame 1L01 rrrr This LLC frame is used for synchronous transmission Address Fields The address fields in FDDI can be either 16 or 48 bits in length Stations with 16 bit addresses however must be able to function in rings with 48 bit ad dresses This means they must be able to repeat 48 bit addresses and to react correctly on receiving claim token and broadcast frames with 48 bit addresses Stations with 48 bit addresses must have also a fully functional 16 bit address and be able to recognize other 16 bit addresses F T 48 bit address _ 16 bit address ad p ap uP 46 bits iy p ap 15 bits iy Universal local bit Individual group bit Figure 9 7 The address field in FDDI TROUBLESHOOTING LocaL AREA NETWORKS 273 FDDI The first bit of a destination address indicates whether the destination is an individual address bit 1 0 or a group address Bit 1 1 bit 1 of the source address is always set to 0 however A group address consisting entirely of 1s is a broadcast address used to send a frame to every station in the ring An address consisting entirely
37. re tilter Stripped Frames 0 0 0 0 1000 0 Pre filter Other Frames 0 0 0 0 100 0 Pre filter Claims 0 0 0 0 700 0 J Pre filter Beacons 0 0 0 700 0 B Pre filter Void Frames 0 0 0 0 100 0 oO Pre filter Bad FCS 0 0 0 0 100 0 Pre filter Violations 0 0 100 0 Pre filter E Bit Set 0 0 0 0 100 0 Pre filter Long FDU 0 0 0 0 100 0 0 000 0 000 0 000 n a 100 0 Post ilter Broadcasts 124 8 124 291 100 1 Post filter Multicasts 0 0 4 40 100 0 Capture Buffer verwrites 0 0 0 0 0 0 Ready FDDI Off Line OM Fiber Das A1450 B450 Figure 9 10 Trend measurements and correlation analysis in FDDI networks using the Agilent Advisor protocol analyzer be localized occur periodically or can at least be reproduced then the trouble shooting process continues with the network component nearest to the problem If the problem source cannot be detected there the range of analysis is succes sively expanded For example if the problems are found to be related to a single network node the next step is to analyze the station s software and hardware components If no fault is found the examination progresses to the patch cables the connectors the wall jack the concentrator and the cabling If the problem cannot be localized at all or if problems that were thought to have been isolated cannot be pinpointed then the only way to find the source of the problem is through systematic segmentation of the network To do this divide the ring physically in
38. recover from the effects of noise S_Min F_Max L_Max Token Holding Timer THT The THT controls the amount of time during which a station may transmit data packets Once the station has obtained a token it may transmit until this timer expires and the Token Rotation Timer TRT remains below the node s priority level T_Pri When a node receives a token it resets its THT with the value remaining in the TRT see the following Timer Valid Transmission TVX The TVX allows a node to recover from an error situation TVX gt max D_Max F_Max Token_Time F_Max S_Min and TVX gt 2 35 ms The default value of TVX is at least 62 500 symbol times or 2 50 ms Token Rotation Timer TRT The TRT controls the ring timing during normal operation When this timer runs out or when an early token is received a token that arrives at a node before the TRT runs out the TRT is initialized with the value currently valid for the TROUBLESHOOTING LocaL AREA NETWORKS 9 276 FDDI operative Token Rotation Timer the T_Opr In the former case the Late counter Late_Ct is also increased by one T_Opr is between the T_Min and T_Max values for the ring and is set upon completion of a claim token process see the following Due to the nature of the token passing protocol it may take up to a whole T_Opr period for a station to receive a token If a station offers a guaranteed T_Resp then T_Opr should be set to 0 5 T_Resp If a station
39. reted as noise e Invalid signals e Elasticity buffer errors while receiving e A mixed symbol pair such as a control indicator paired with a data symbol e Ann R S or T symbol or a symbol pair containing at least one of these symbols received while the line state is not ILS or ALS e Reception of an I n R S or T symbol while the clock detect function a mechanism that monitors clock synchronization reports a synchroniza tion error 9 1 1 2 The Elasticity Buffer Differences are bound to occur between a receiving node s internal oscillator and the clock rate of the incoming bit stream due to the transmission medium and to tolerance limits in network components If the transmission rate of a given station is significantly lower than the incoming data rate data could be lost To prevent this each station has an elasticity buffer to compensate for differences of up to 4 5 bits or 0 01 percent The frequency of the local oscillator must meet the following specifications Nominal frequency 125 MHz 0 005 50 ppm Phase jitter at 20 kHz lt 8 degrees Harmonic content at 125 02 MHz lt 20 dB Nominal code bit time 8 0 ns Nominal symbol time 40 0 ns 9 1 1 3 Smoothing The smoothing function ensures that the preamble of an FDDI frame is not lost in the process of passing through a number of elasticity buffers This function removes surplus symbols from oversized preambles and adds them to under sized preambles The smoot
40. river software An oversized frame is any frame of more than 4 500 bytes Its LE bit is set to 1 295 Symptom Cause Symptom Cause Symptom Cause TROUBLESHOOTING LocaL AREA NETWORKS 9 FDDI Token Rotation Time is Too Long Problems with station configuration or cabling Similar to statistics on capacity use the TRT is also an indicator of ring performance It should lie below the TTRT negotiated dur ing the claim process If the TRT regularly goes over the negoti ated TTRT this could be an indication of incorrect station con figuration or of problems in cables or connectors Invalid Frames Violation Frames Station detects invalid symbols When a station detects invalid symbols it reports this in the next valid frame it transmits The frame with the error message is not the frame that contains the coding violation or error The error domain is upstream from the station that reports the violation Check all the components in the upstream transmission path in cluding concentrators cables connectors and the interface card in the neighboring station until you locate the source of the error Interface Overflow Wedged Interface Bursts of small packets that overflow the queue Wedged interface ports are a common problem In these cases the input output queue exceeds the maximum value supported by the router port The solution is either to increase the queue size or to reload the router TROUBLE
41. rly or intermittently e Are the symptoms related to certain applications or do they affect all network operations e Do the symptoms correlate to other activities in the network e When was the first occurrence of the symptom e Was there any change in any hardware or software network component e Has anyone connected or disconnected a PC laptop or desktop or any other component to or from the network e Has anyone installed an interface card in a computer e Has anyone stepped on a cable e Has any maintenance work been performed in the building recently by a telephone company or building maintenance personnel for example e Has anyone including cleaning personnel moved any equipment or furniture 9 2 2 Starting the Troubleshooting Procedure Troubleshooting in FDDI LANs is primarily performed using cable testers for optical fiber and twisted pair copper wire protocol analyzers and special FDDI ring management software to track and display SMT functions FDDI has several self diagnosis functions that enable it to recover from a number of critical states on its own Key requirements for successful troubleshooting in an FDDI network include a detailed understanding of its operational processes and of the SMT functions If the ring is still functional the first step in the troubleshooting procedure involves using a protocol analyzer to determine the main operating statistics of the network These statistics include ring load as
42. rol destination ad dress source address information frame check sequence end of frame se quence ending delimiter and frame status PA SD FC DA SA INFO FCS ED FS SFS Covered by FCS EFS SFS iv Start of Frame Sequence INFO Information 0 or more symbol pairs PAn Preamble 16 or more symbols POS wissen Frame Check Sequence 8 symbols SD Starting Delimiter 2 symbols EF Seriese End of Frame Sequence FO iisi Frame Control 2 symbols EDaren Ending Delimiter 1 symbol DA Destination Address 4 or 12 symbols F Shiai Frame Status 3 or more symbols SA Source Address 4 or 12 symbols Figure 9 5 Format of an FDDI frame 9 1 2 1 Token and Data Packet Fields Preamble PA A preamble consists of at least 16 idle symbols although the length can vary during circulation through the ring due to differences in nominal frequency and to smoothing and elasticity buffering Frames with a preamble of fewer than 12 symbols are not copied into the destination station s receive buffer Starting Delimiter SD Every frame including tokens begins with this field which consists of the symbol sequence JK TROUBLESHOOTING LocaL AREA NETWORKS FDDI 9 271 Frame Control FC C LFF Z ZZZ Cases Class bit bios Address length bit Fra Format bits ZZZZ Control bits Figure 9 6 The frame control field in FDDI The frame control field identifies the frame type It consists of frame class bits ad
43. t of all the stations connections protocols and services affected by the problems observed To do this measure the current parameters in the various network segments and compare the results with statistics gathered during normal operation This involves recording and analyzing throughput and perfor mance parameters of network nodes protocols and applications as well as reviewing log files that contain the operating statistics on all bridges in the network The log files provide bridge statistics such as CPU capacity use port capacity use buffer capacity use and error rates To measure the response times of connections across bridges send echo frames across the bridges from different network segments Long term response time measurement statistics can be especially useful in diagnosing intermittent problems Based on the results of the measurements the range of potential sources of error can usually be narrowed down to specific components 9 2 6 2 Symptoms and Causes of Bridge Problems The symptoms for most bridge problems in FDDI networks differ only slightly from those in Ethernet or Token Ring networks As described in the section on Ethernet bridges the most common difficulties are throughput problems incor rectly configured filter settings bridge buffer overflow and faulty address tables Problem characteristics of FDDI networks include bridge ports config ured with duplicate FDDI addresses and incorrect frame length settings Br
44. timing errors are estimated at a maximum of 4 bits Consequently SD_Max lt 592 ns 4 80 80 756 ns P_Max The number of physical FDDI interfaces in the ring The default value is 1 000 which corresponds to 500 dual attachment stations D_Max The maximum transmission delay of a starting delimiter sequence when no noise is present Thus a combination of 1 000 FDDI interfaces a ring length of 100 km and a signal propagation speed of 5 085 ns km yields D Max P_Max x SD_Max 2 x 100 x 5085 1 773 ms The default value for D_Max should be less than 1 773 ms the specification calls for 1 617 ms 9 1 2 The FDDI Data Format There are two types of frames in FDDI networks tokens and data packets A token is 3 bytes long and consists of a starting delimiter a frame control field and an ending delimiter A token is a special frame that is passed from station to TROUBLESHOOTING LocaL AREA NETWORKS 9 270 FDDI station and controls access to the LAN medium If a given station receives a valid token but cannot forward it for some reason such as a ring timing error then the station issues a new token PAn Preamble 16 or more symbols gin Frame Control 2 symbols SD suku Starting Delimiter 2 symbols E Digs Ending Delimiter 2 symbols Figure 9 4 Structure of an FDDI token All other data packets can have lengths of 12 to 4 500 bytes and consist of the following fields preamble starting delimiter frame cont
45. tions transmit SRFs to inform other components of changes in their configuration The presence of a large number of status report frames may indicate problems in the FDDI ring Use a protocol analyzer or the ring management system to collect and analyze the SRFs If they do not indicate any unusual conditions transmit SIFs to poll stations on their status Keep in mind that the error counters maintained by each node count only frames that end with a valid ending delimiter Frames that end in Idle symbols or invalid characters can only be detected using a proto col analyzer High Numbers of Claim Frames Expired TVX or TRT The station has not received a valid token or data packet for over 2 5 ms This may be due to a high BER which may in turn result from cable or connector problems defective FDDI ports or prob lems with optical bypass switches High Checksum Error Rates FCS Errors Defective cable defective FDDI interface card dust dirt or finger prints on the MIC connector FDDI Frames with the Error Bit Set Defective cable defective FDDI interface card The error domain is directly upstream from the station that sets the E bit in the frames Check all the components in the upstream transmission path including concentrators cables connectors and the interface card in the neighboring station until you locate the source of the error Oversized Data Packets Length Error Bit Set Problems with the interface card or d
46. tive HQ or QH symbol pairs are received and ends as soon as any other symbol pair is received Halt Line State HLS HLS is entered when H symbols are transmitted continuously while a connec tion is being set up This state is detected as soon as 16 or 17 H symbols are received and exited when any other symbol is received or when the signal is lost Idle Line State ILS The ILS characterized by a continuous stream of I symbols is entered while a connection is being set up and during the transmission pauses between data packets The state is recognized when four or five consecutive I symbols are received The elasticity buffer see the following may increase this value by up to 11 bits ILS is exited when any other symbol is received or when the signal is lost Active Line State ALS ALS indicates that the incoming bit stream consists of valid FDDI frames meaning that the nearest upstream neighbor has an active connection to the TROUBLESHOOTING LocaL AREA NETWORKS 9 268 FDDI ring This state is entered once a starting delimiter is received ALS is exited upon receipt of any symbol other than I n R S or T n any data symbol upon loss of a valid signal or on entering ILS Noise Line State NLS This line state indicates that the incoming signals are distorted by noise and that the physical connection is faulty This state is entered upon receipt of 16 or 17 consecutive invalid symbols The following events are interp
47. to two rings determine which of these still shows the error condition divide that ring into two and so on until the error is localized This method causes considerable disruption in network operation and is therefore applied only as a last resort when the problem itself severely impairs normal network operation TROUBLESHOOTING LocaL AREA NETWORKS 9 288 FDDI If the symptoms occur intermittently long term measurements are necessary These must be performed continuously until the basic network operating statis tics have been measured during the occurrence of the fault This information usually provides the first clue to the error source Furthermore it is essential to log the exact time of intermittent error events Later this information can be used to find temporal correlations with other events in the network or on a given node such as backups the start of specific applications connections through routers access to the Internet users working hours or other possible factors If this does not help to track down the error you may have to resort to the segmentation method Depending on which causes the least inconvenience to users you can either systematically disable network functions and applications or physically separate concentrators These methods usually lead to the error source 9 2 3 Error Symptoms in FDDI The most common symptoms of problems in FDDI networks are ring wrapping a large number of claim or beacon frames
48. vators or produc tion environments These problems are discussed in detail in the chapter on cabling Two factors that must be mentioned with specific reference to FDDI however are the maximum bit error rate BER permitted between two FDDI stations and the optical bypass function The BER due to signal repetition must not exceed 2 5x10 If the signal power exceeds the minimum requirement by 2 dB then the BER must not exceed 1x107 At the receive port a signal power of 31 dBm or more must be recognized as valid within 100 us Another possible source of errors is the optical bypass function in ring nodes This function isolates a node upon failure so that the double ring architecture of the network is maintained Without the bypass function failure of a node causes the ring to wrap which means it doubles back on itself and is reconfigured as a single ring The following specifications are defined for bypass functions used in FDDI Min Max Units Attenuation input output 0 0 2 5 dB Optical switching time 15 ms Station switching time 25 ms TROUBLESHOOTING LocaL AREA NETWORKS 9 291 FDDI The optical switching time is the time during which the primary and secondary optical signal is interrupted during the switching process measured from the time the signal drops more than 1 5 dB below the original signal level S1 to the time the signal recovers above S1 1 5 dB The station switching time is the duration of the
Download Pdf Manuals
Related Search