Best way to do containerised healthchecks on workers

[danielloader]

Hey been working with chancy for a few days and got as far as a working prototype, but now I'm wondering how in docker/kubernetes how I can check a worker is still working - either as a http payload or exec/command running in the container.

Is there a preferred way to check the worker to see if the heartbeat is still ticking and the worker hasn't failed connection to the database to get more jobs from outside the python application?

[TkTech]

That's a good Q! In our use case we do external orchestration, so the heartbeat table and the metrics table for throughput are enough. There is a limited API plugin which is used by the dashboard and so is mostly undocumented as its quite in flux. v0.26 has a complete rework of the dashboard and the API so this may be a good time to make it stable.

We could add a healthcheck endpoint to this API easily enough. What are you expecting to see in the result?

[danielloader]

Great news, a simple handler issuing a http 200 when the heartbeat is still good would be plenty.

Doesn't need a body or details.

[TkTech]

In #58, there's a new unauthenticated /api/v1/health endpoint which returns 200 if the worker is still alive and kicking.

[ppsmilesi]

Nice ! I personnally created my own healthcheck plugin to support this. Would love to have this part of chancy !