neutron-service is throwing an excessive amount of warnings

[Joe Gordon]

http://logs.openstack.org/58/136158/9/check/check-tempest-dsvm-neutron-full/abea49f/logs/screen-q-svc.txt.gz?level=TRACE

Is full of a lot of repetitive warnings, and it is unclear if something may be going wrong or not.

[Eugene Nikanorov (enikanorov)]

That's a normal case for tempest tests, especially API tests which perform operations too quickly.

I don't think we need to do anything special with it.

[Joe Gordon (jogo)]

I disagree, neutron should not be logging warnings for things that are not possibly issues. And if something is going wrong so often that we get this many warnings it probably should be fixed.

[Eugene Nikanorov (enikanorov)]

In each case it might be an issue, it might be not.
API tests are special case where client doesn't care about consistent state between the backend (agents) and the database.
That consistent state becomes broken and that's why it is logged at warning level.

I think in real life scenarios it makes sense to detect such cases at logging level that is visible by default (e.g. INFO+).
We can lower it to info, but amount of logs will be the same.
What do you think?

[Joe Gordon (jogo)]

Lowering the log severity in a case by case basis sounds like a good approach to me. What do you mean the API tests don't care about consistent state between agents and the database? When is it ever ok for the backends to get out of sync with the databse?

[Eugene Nikanorov (enikanorov)]

For instance, tempest API tests used to create and then immediately delete an object, or create-update-delete.
Backend (agent side) might not keep up with frontend state changes, and thus it logs warnings that it could not find objects that should be there.
I think that's ok to have those messages at warning level, because the case is not practical.
In real life scenarios you would care about backend state too, but if backend detects that frontend went out of sync - it deserves visible logging level IMO

[Joe Gordon (jogo)]

http://logs.openstack.org/04/136504/15/experimental/check-tempest-dsvm-neutron-aiopcpu/5027134/logs/screen-q-agt.txt.gz?level=TRACE
http://logs.openstack.org/04/136504/15/experimental/check-tempest-dsvm-neutron-aiopcpu/5027134/logs/screen-q-dhcp.txt.gz?level=TRACE
http://logs.openstack.org/04/136504/15/experimental/check-tempest-dsvm-neutron-aiopcpu/5027134/logs/screen-q-svc.txt.gz?level=TRACE

These are logs from a passing gate job. These logs are full of so many warnings errors and stacktraces that it is really hard to spot real issues in them

[Armando Migliaccio (armando-migliaccio)]

This bug is > 365 days without activity. We are unsetting assignee and milestone and setting status to Incomplete in order to allow its expiry in 60 days.

If the bug is still valid, then update the bug status.

[Launchpad Janitor (janitor)]

[Expired for neutron because there has been no activity for 60 days.]