Comment 33 for bug 1927868

Revision history for this message
Edward Hope-Morley (hopem) wrote :

I've had a go at deploying Train and upgrading Neutron to latest Ussuri and I see the same issue. Looking closer what I see is that post-upgrade Neutron l3-agent has not spawned any keepalived processes hence why no router goes active. When the agent is restarted it would normally receive two router updates; first one to spawn_state_change_monitor and a second to spawn keepalived. In my non-working nodes the second router update is never received by the l3-agent. Here is an example of a working agent https://pastebin.ubuntu.com/p/PFb594wkhB vs. a not working https://pastebin.ubuntu.com/p/MtDNrXmvZB/.

I tested restarted all agents and this did not fix things. I then rebooted one of my upgraded nodes and it resolved the issue for that node i.e. two updates received and both spawned then router goes active. I also noticed that on a non-rebooted node, following ovs agent restart I see https://pastebin.ubuntu.com/p/2n4KxBv8S2/ which again is not resolved by an agent restart and is fixed by the node reboot. This latter issue is described on old bugs e.g. https://bugs.launchpad.net/neutron/+bug/1625305