OPERATIONAL DEFECT DATABASE
...
...
After RP failover bundlemgr process stuck in initialization. Which causes *all of the bundle interfaces* go down. Bundle members does not come up due to multiple reasons. 1. *All* bundle interfaces down. # show interface summary -> Shows all bundle interfaces down. # show bundle -> Shows details of all bundles with members down. -> Some members stuck in "BFD link dows" -> Some members stuck in "Too many bundle member links configured" -> Some members stuck in "LACPDUs not received from partner" -> Some members stuck in "Replication" 2. Check the bundlemgr pending operations: # show bundle infrastructure in-progress ma -> Shows pending operations for "BFD SCN pending" for some of bundle-members. 3. Check the bundlemgr process state: # show bundle infrastructure database ma -> Shows BMM (0/RP0/CPU0) - Active RP, "Process role" as "Active", which is supposed to be "Primary Active". All of above symptoms shall occur.
When one of the bundle is misconfigured, i.e. bundle has bfd configuration, but does not have ipv4 address configured. 1. Check the bundle configurations for members shows operation is pending. # show run interface 2. Check if bfd is configured on bundles and ipv4 address *is not* configured. # show run interface bundle-ether [N] -> Shows "bfd address-family ipv4 *" configurations but *does not* have "ipv4 address " configuration.
How to avoid this problem: --------------------------------- Make sure bundle interface is configured with ipv4 address if bfd configuration is present. e.g. all of the following config (but not limited to) considered to be necessary to avoid this problem on both devices (local and remote): (config)#interface Bundle-Ether (config-if)#bfd address-family ipv4 multiplier (config-if)#bfd address-family ipv4 destination (config-if)#bfd address-family ipv4 fast-detect (config-if)#ipv4 address / How to recover: -------------------- 1. Configure the ipv4 address on bundle, if the bfd configuration is present. (config)#interface bundle-ether (config-if)#ipv4 address / This local-bundle-intf-ip-addr (e.g. a.b.c.d) should be same as remote device's bfd config - "bfd address-family ipv4 destination a.b.c.d" 2. Or remove the bfd configuration if not required on bundle interface from both devices (local and remote). (config)#interface bundle-ether (config-if)#no bfd address-family ipv4 timers start (config-if)#no bfd address-family ipv4 timers nbr-unconfig (config-if)#no bfd address-family ipv4 multiplier (config-if)#no bfd address-family ipv4 destination (config-if)#no bfd address-family ipv4 fast-detect
If the IP is not configured on bundle (which has bfd configuration), bundlemgr process tries to create bfd session and waits for SCN (State Change Notification) from bfd-server. In this case, bfd server wont send SCN as the state is Unknown. Waiting for SCN causes bundle process to wait indefinitely and holds the process initialization. This is a baseline defect, occurs due to misconfig and timing condition.
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.
