Open nazariig opened 3 years ago
@prsunny please have a look
From my analysis and the logs, this subnet route delete is triggered by FRR zebra. Verified on both 201911 and 201811 and below are the observations:
Dec 12 00:27:17.557194 str-sn3800-01 NOTICE swss#orchagent: :- updatePortOperStatus: Port Ethernet72 oper state set from up to down
Dec 12 00:27:17.557410 str-sn3800-01 NOTICE swss#orchagent: :- setHostIntfsOperStatus: Set operation status DOWN to host interface Ethernet72
Dec 12 00:27:17.561697 str-sn3800-01 DEBUG bgp#fpmsyncd: :- onRouteMsg: Receive new route message dest ip prefix: fe80::/64
Dec 12 00:27:17.561697 str-sn3800-01 DEBUG bgp#fpmsyncd: :- onRouteMsg: RouteTable set msg: fe80::/64 :: eth0
Dec 12 00:27:17.561697 str-sn3800-01 DEBUG bgp#fpmsyncd: :- onRouteMsg: Receive new route message dest ip prefix: 72.1.1.0/24
2\. [201811 - Quagga] fpmsyncd is _not_ getting the route msgs from zebra during oper status change.
@prsunny so what's the plan? Do we need a bug for FRR?
Potential fix: https://github.com/FRRouting/frr/pull/7745
Description Local subnet routes for IPv4/IPv6 are being removed on link operational state down. This causes packets to be looped thru the default route if any.
Steps to reproduce the issue:
Describe the results you received:
Link operational state is up:
Kernel:
Debugger:
Logs:
Link operational state is down:
Kernel:
Debugger:
Logs:
Describe the results you expected: Local subnet routes should not be removed on link operational state down
Additional information you deem important (e.g. issue happens only occasionally): Switch configuration:
Netlink reference:
Output of
show version
:Attach debug file
sudo generate_dump
: