Symptom
Routing protocols that are peering using loopback addresses flap after FO.
Conditions
Interface manager process crashes immediately after FO but before sending go-active notifications to the ipv4 and ipv6 management agent processes for virtual interfaces.
Workaround
Restart ipv4_ma and ipv6_ma on the newly active RP within 5 minutes of the ifmgr crash.
Further Problem Description
Interface manager process crashes immediately after FO but before sending go-active notifications to the ipv4 and ipv6 management agent processes for virtual interfaces. These notifications are critical for these management agents on the newly active RP to assume ownership of the virtrual interfaces such as loopbacks. Upon restart, the IM doesn't send these notifications again (per design) and the management agents don't take care of this situation.
Note that the IM crash immediately after FO will be a rare situation.
