Comment 8 for bug 1979089

Revision history for this message
Maximilian Stinsky (mstinsky) wrote :

Greetings,

I think we ran into a similar issue if not even the same.
In our prod environment we had a l3 agent stuck trying to start a metadata haproxy for a router which already got removed. This seemed to even get the l3 agent stuck to not be able to update any states for new router anymore, so all new routers got into an unknown state.

A restart of the agent seemed to fix this and it did not try to create the haproxy after that anymore.

We are running neutron wallaby (18.5.0).