search

ansible / eda-server

Event Driven Ansible for AAP
Apache License 2.0
59 stars 35 forks source link

chore: support workers startup order with podman-compose #957

Closed Alex-Izquierdo closed 2 weeks ago

Alex-Izquierdo commented 2 weeks ago

We rely on the healthcheck and depends_on::condition in our docker-compose files to ensure the correct startup order. Since podman-compose doesn't support it, add the wait_for_migrations command before the rq workers to ensure they are executed after the api pod and avoid errors like:

[eda-default-worker]    | django.db.utils.ProgrammingError: relation "core_activation_request_queue" does not exist
[eda-default-worker]    | LINE 1: ...re_activation_request_queue"."activation_id" FROM "core_acti...
Alex-Izquierdo commented 2 weeks ago

It seems the e2e fails because the import project task reach the timeout. I have seen other failures today, I feel CI is having today a bad network performance.

Alex-Izquierdo commented 2 weeks ago

The multinode tests always fails with the same error related with project import. I have no idea why and it seems the change in the PR has not relation with it, but it fails. I need to investigate further why is failing.

Alex-Izquierdo commented 2 weeks ago

It is a performance issue, increasing the timeout in the test suite solves the problem: https://github.com/ansible/eda-server/actions/runs/9764139814/job/26951732995