- Customer reported that bgp peers went down after upgrade from 23.4R2-S3.9 to 24.2R1-S1.9 - Customer reported that only IPV4 peers went down while IPV6 peers remain in established state - Customer has firewall policy filter applied. - When downgraded to 23.4R3-S3.9 with the exact same config, the peers gets into established state automatically. - Can ping to the peer( ping is woking fine) run show bgp summary Threading mode: BGP I/O Default eBGP mode: advertise - accept, receive - accept Groups: 2 Peers: 4 Down peers: 2 Table Tot Paths Act Paths Suppressed History Damp State Pending inet.0 0 0 0 0 0 0 inet6.0 2 1 0 0 0 0 Peer AS InPkt OutPkt OutQ Flaps Last Up/Dwn State|#Active/Received/Accepted/Damped... 96.109.0.25 777 0 0 0 0 2:14:20 Active 96.109.0.29 777 0 0 0 0 2:14:20 Active 2001:558:96:125::1 777 298 298 0 0 2:13:21 Establ inet6.0: 1/1/1/0 2001:558:96:129::1 777 296 298 0 0 2:13:19 Establ inet6.0: 0/1/1/0
Trigger is upgrading from 23.4R2-S3.9 to 24.2R1-S1.9
Starting 24.2 release "set chassis loopback-firewall-optimization" CLI is supported via RLI 56582
* Behaviour reported BGP down is seen only when CLI knob is enabled in 24.2 release.
* BGP packets are dropped because of deny-all term in the lo0 filter.
* Upon capturing the dropped BGP packets, they are reaching with ttl as 1.
* For ttl-1 BGP packets to get accepted when "loopback-firewall-optimization" is enabled, we need to add extra ttl-1 term.
* Later on, BGP is coming up as expected and no issue is seen.
In issue state when we delete the "set chassis loopback-firewall-optimization" CLI, issue is NOT seen. BGP is up in 24.2 release.
TTL-1 BGP are dropped due to "Loopback-firewall-optimization" enabled without ttl-1 term. To accept TTL-1 BGP packets when loopback-firewall-optimization is enabled, it is necessary to add an additional term for TTL-1 in the lo0 filter.
loopback-firewall-optimization