Cisco Bug: CSCvu69336 - BGP recursing to BGP is causing traffic loss due to hardware programming backwalk
Aug 26, 2020
- Cisco ASR 9000 Series Aggregation Services Routers
Known Affected Releases
Symptom: customer is not using any IGP. Loopback and label information is carried in BGP ipv4-LU. Customer then peers with remote PE/RR over IPv4 using the loopback/label information. When the LU next hop best path changes, it triggers a hardware backwalk of all prefixes using that next hop. With 500k prefixes on asr9k, we are seeing a 26 second delay in programming the line card hardware. The routing switches immediately, but traffic is sent with wrong label and dropped at destination. After the hardware programming is complete and label is correct, traffic resumes as normal. A second impact is seen if that best path is restored and the best path again changes to the original. Conditions: customer is not using any IGP. BGP IPv4 uni is recursing to a next hop learned from BPG IPv4 LU. The best path for the LU next hop changes due to link failure or cost out. If a non best path link fails or is costed out, then no impact is seen. In case of multipath, even if the specific flow hashes to the backup path, when the primary path fails ALL prefixes to that next hop are impacted, even if they hash to the backup link.
Bug details contain sensitive information and therefore require a Cisco.com account to be viewed.
Bug Details Include
- Full Description (including symptoms, conditions and workarounds)
- Known Fixed Releases
- Related Community Discussions
- Number of Related Support Cases