
OPERATIONAL DEFECT DATABASE
...

...
The main issue we have found in the lab is having a long packet loss duration when configuring BoB and having the Stand-By RP ready. Using Dual N560-4-RSP4E. IOS XR 6.6.3
Bundle Ethernet configured using member on two different IM cards, having 50msec x 3 timers Condition occurs when a member link of the Bundle is shutdown at the peer router and then unshutdown: The BFD session on the member link starts and gets into dampening state. At this point the traffic using this member link drops for about 7 seconds. When all conditions get steady, the BFD session on the affected member link on the Stand-By RP does not sync to the established timers. Same test when using a single RP board, then the traffic is not lost at the time of unshutting the interface. RP/0/RP0/CPU0:NCS_560#show bfd session Thu Jul 23 12:31:10.688 UTC Interface Dest Addr Local det time(int*mult) State Echo Async H/W NPU ------------------- --------------- ---------------- ---------------- ---------- Te0/3/0/0 192.168.0.1 0s(0s*0) 150ms(50ms*3) UP Yes 0/RP0/CPU0 Te0/2/0/3 192.168.0.1 0s(0s*0) 150ms(50ms*3) UP Yes 0/RP0/CPU0 Te0/3/0/0 192.168.0.1 0s(0s*0) 150ms(50ms*3) UP Yes 0/RP1/CPU0 Te0/2/0/3 192.168.0.1 0s(0s*0) 6s UP Yes 0/RP1/CPU0 BE100 192.168.0.1 n/a n/a UP No n/a
N/A
NCS560-4 Bundle Ethernet operating correctly to Third-party device 7750SR TiMOS-C-16.0.R5 Bundle configured using two (02) different IM cards, ports Ten0/3/0/0 and Ten0/2/0/3 BFD session when configured: 50msec x 3 When 7750SR device shutdown one member link -> Packet loss duration measure with Ixia traffic generator = 20 msec. Result OK When 7750SR device unshut same member link -> Packet loss duration depends if the Bundle has configured BFD or not: - If no BFD configured -> Packet loss duration measured with Ixia traffic generator ~= 1 msec. Result OK - If BFD configured -> Packet loss duration measured with Ixia traffic generator ~= 7 seconds (7000 msec). Result NOT EXPECTED A second test: shutting down stand-by RP and unshutting the member on the far end to add it to the bundle. This result Bundle Ethernet not causing more than a few milliseconds of packet loss. Third test: shutting down BFD agent only on stand-by RP and unshutting the member on the far end to add it to the bundle. This result Bundle Ethernet not causing more than a few milliseconds of packet loss. Techzone with part of the analysis: https://techzone.cisco.com/t5/IOS-XR-L3FIB-Eng-Discussion/NCS560-ATT-MX-6-6-3-Bundle-with-BoB-has-long-packet-loss/m-p/1568563#M8585 When analysing the logs, the BFD dampening is ending on different times on the two RPs; first on Stand-by RP1 and later on active RP0. RP0: ===== RP/0/RP0/CPU0:Aug 13 10:42:54.201 UTC: bfd_agent[280]: %L2-BFD-6-SESSION_DAMPENING_ON : Session to neighbor 192.168.0.1 on interface TenGigE0/2/0/3 entered Dampened state (initial: 16000 ms,secondary: 20000 ms,maximum: 600000 ms). RP/0/RP0/CPU0:Aug 13 10:43:10.041 UTC: bfd_agent[280]: %L2-BFD-6-SESSION_DAMPENING_OFF : Session to neighbor 192.168.0.1 on interface TenGigE0/2/0/3 moved out of Dampened state. RP1: ==== RP/0/RP1/CPU0:Aug 13 10:42:54.212 UTC: bfd_agent[456]: %L2-BFD-6-SESSION_DAMPENING_ON : Session to neighbor 192.168.0.1 on interface TenGigE0/2/0/3 entered Dampened state (initial: 16000 ms,secondary: 20000 ms,maximum: 600000 ms). RP/0/RP1/CPU0:Aug 13 10:42:56.212 UTC: bfd_agent[456]: %L2-BFD-6-SESSION_DAMPENING_OFF : Session to neighbor 192.168.0.1 on interface TenGigE0/2/0/3 moved out of Dampened state. The main issue we have found in the lab is having a long packet loss duration when configuring BoB and having the Stand-By RP ready. Using Dual N560-4-RSP4E. IOS XR 6.6.3 Bundle Ethernet configured using a member on two different IM cards, having 50msec x 3 timers The condition occurs when a member link of the Bundle is shutdown at the peer router and then unshutdown: The BFD session on the member link starts and gets into dampening state. At this point, the traffic using this member link drops for about 7 seconds. When all conditions get steady, the BFD session on the affected member link on the Stand-By RP does not sync to the established timers. The same test when using a single RP board, then the traffic is not lost at the time of unshutting the interface.
Click on a version to see all relevant bugs
Cisco Integration
Learn more about where this data comes from
Bug Scrub Advisor
Streamline upgrades with automated vendor bug scrubs
BugZero Enterprise
Wish you caught this bug sooner? Get proactive today.