Develop Invenio Helm Charts

[Samk13]

Goals

Bring the current helm-invenio repository to a working/usable state for a Kubernetes deployment.

Resources

KTH has assigned a developer to focus on this project for the first quarter of 2024.

[lnielsen]

@Samk13 Can you provide a quick update on this issue? Not an issue if nothing has been done, just want to check up on if there's progress on it.

[Samk13]

@lnielsen: The one in charge of this task is @lindhe, who is currently on vacation. There are some current obstacles, but more specific information should be available upon their return.
The current status is operational, however, it will not function on a local Minikube setup and requires a specialized configuration.
There are manual interventions required for health probes that need to be addressed on setup, but other than that, it is functional.

[lindhe]

Like Sam says, by far the most important thing to get in order is inveniosoftware/helm-invenio#104 so one does not have to patch to remove the health probes every time one installs/updates the chart. There has been some discussions there with suggestions and there has been new image versions released since I tried last, so I should really test it again to try and verify exactly what the current behavior is so I can continue the discussion in that Issue.

In inveniosoftware/helm-invenio#106 I've raised the issue of the chart being hard to deploy in local environments (like minikube). I've had some discussions on Discord with the maintainers, and I think they would like to see some changes to make that possible. I think that is probably not too much work to fix.

If I recall correctly, a major blocker for this was due to using access mode ReadWriteMany on the volume (which is rarely supported on simple single-node systems). But it should be possible to use ReadWriteOnce instead, as long as everything is scheduled on the same node.

The other major blocker was that Invenio and the bundled Nginx proxy seems to be very particular about the HTTPS settings. That's a good thing for ensuring security when running in prod, but in a local dev environment it's problematic if it requires a valid TLS certificate and publicly routable FQDNs. This is probably more tedious work to fix, trying to go through the Nginx config and possibly Invenio configuration variables too.