Description of problem:
The server often disconnects from the client because of probes are not being replied to. It's not because of the load on the server since server is the one waiting for the reply and it happens in both NB and SB dbs and only to connections to the neutron server.
2022-09-11T00:01:36.310Z|132111|reconnect|ERR|tcp:10.0.37.88:39878: no response to inactivity probe after 361 seconds, disconnecting
2022-09-11T00:02:09.779Z|132127|reconnect|ERR|tcp:10.0.37.88:45798: no response to inactivity probe after 360 seconds, disconnecting
2022-09-11T00:02:48.809Z|132148|reconnect|ERR|tcp:10.0.37.88:45630: no response to inactivity probe after 361 seconds, disconnecting
2022-09-11T00:05:16.381Z|132186|reconnect|ERR|tcp:10.0.37.88:50510: no response to inactivity probe after 436 seconds, disconnecting
2022-09-11T00:07:14.981Z|132238|reconnect|ERR|tcp:10.0.37.88:58688: no response to inactivity probe after 370 seconds, disconnecting
2022-09-11T00:07:38.006Z|132245|reconnect|ERR|tcp:10.0.37.88:33770: no response to inactivity probe after 360 seconds, disconnecting
2022-09-11T00:08:25.866Z|132279|reconnect|ERR|tcp:10.0.37.88:37142: no response to inactivity probe after 365 seconds, disconnecting
2022-09-11T00:08:50.498Z|132280|reconnect|ERR|tcp:10.0.37.88:38800: no response to inactivity probe after 360 seconds, disconnecting
2022-09-11T00:09:31.489Z|132281|reconnect|ERR|tcp:10.0.37.88:37446: no response to inactivity probe after 361 seconds, disconnecting
2022-09-11T00:11:16.994Z|132321|reconnect|ERR|tcp:10.0.37.88:44330: no response to inactivity probe after 361 seconds, disconnecting
It doesn't happen when neutron_api container is downgraded to 16.1.7 version. There is fairly large number of TRY_AGAIN messages.
[ctrl-net-d-01.infra.prod.upshift.rdu2.redhat.com] [01:08:21 AM]
[root@ctrl-net-d-01 jlibosva_logs]# grep TRY_AGAIN server.log.1 -c
2069103
Version-Release number of selected component (if applicable):
16.1.8
How reproducible:
Always
Steps to Reproduce:
1. Start neutron server
2.
3.
Actual results:
Expected results:
Additional info:
Updating the inactivity probe to 0 in the neutron ovn config file under the [ovn] section on all 3 controllers and restarting neutron_api worked around this issue for now.
```
[root@controller-0 ~]# grep "ovsdb_probe_interval" /var/lib/config-data/puppet-generated/neutron/etc/neutron/plugins/ml2/ml2_conf.ini
ovsdb_probe_interval = 0
```