Add variable to disable healthcheck for aio-controlled containers

[crgmz]

Is your feature request related to a problem? Please describe.

N/A

Describe the solution you'd like

Add an env variable like NEXTCLOUD_DISABLE_HEALTHCHECKS to disable the built-in healthchecks for the internal nextcloud/database/apache/redis/etc containers.

Describe alternatives you've considered

I haven't found a way to persistently disable healthcheck for the internal containers that would survive restarts/updates

Additional context

N/A

[szaimen]

Hi, can you please clarify why do you need this?

[crgmz]

Hello, I'm running aio on a resource constrained system (by choice) and sometimes a transient load spike can lead to services being marked as unhealthy and triggering unnecesary alerts but everything goes back to normal once the spike is gone. Since I already have external monitoring in place I thought it would be useful to be able to disable internal healthchecks.

[szaimen]

I see the problem but I fear this is out of scope for now. Maybe you can exclude these containers from your healthceck monitoring instead?

[crgmz]

No problem, I have some workarounds for it. It was more of a nice-to-have, thank you anyways

[diego351]

@crgmz Can you share these workarounds?

[crgmz]

@diego351 @szaimen

The internal containers are created via DockerActionManager.php, so I patch that file to disable healthchecks using standard docker api: $requestBody['Healthcheck'] = ['Test' => ['NONE']];.

That file gets overwritten when AIO updates so I have a cron that reapplies the patch daily, and I also have AIO automatic backups enabled which stop/start containers to run the backup, which makes sure the internal containers start with my patch applied and thus no healthchecks.

That one line patch could easily be gated behind a NEXTCLOUD_DISABLE_HEALTHCHECKS flag, but I haven't taken the time to analyze the potential blast radius so I've settled for my manual patch.