Refs. swh/infra/sysadm-environment#4812

This also unifies the entrypoint.sh to be installed as the other images do.

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#4812

It was the inlined configuration in the k8s-cluster-config. It's now deprecated in favor
of the actual configuration being configured properly in the values files.

Expectedly the secrets part are still stored in the private repository.

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#4812

That way we can pass inline the configuration to the configuration map. And then iterate
over this to deal with more privatey configuration in the next commits.

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#4812

This also ensures we reuse the helper function to add the environment secrets when
needed.

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#4812

Vincent Sellier authored 1 year ago and Antoine R. Dumont committed 1 year ago

Related to swh/infra/sysadm-environment#4812

Vincent Sellier authored 1 year ago and Antoine R. Dumont committed 1 year ago

Refs. swh/infra/sysadm-environment#4812

so it's idempotent.

Refs. swh/infra/sysadm-environment#4812

Refs. swh/infra/sysadm-environment#5077

It now contains the psql tool.

Refs. swh/infra/sysadm-environment#5077

It prevents the actual scheduler migration to work.

Refs. swh/infra/sysadm-environment#5077

This isn't supported by FreeBSD (at least when NATing), and apparently
this causes connections to hang

Refs. swh/infra/sysadm-environment#5048

Images are explicitly tagged and not regenerated, they should only be
pulled on first install.

We use "write-only" tags for our deployed images so there's not much
point pulling them once they're present on the server: they're not going
to change anymore.

There's no point hammering the upstream for images when they're updating
only every month or two.

The queue has been stable around 5k without moving for a while. Let's increase the
workload.

The queue has subsided to a normal size now.

Related to swh/infra/sysadm-environment#5042

It's clearer that way. I missed that before.

It was missing the '-' in the list value.

This allows to declare ingresses which requires some specific setup on specific
endpoints (whitelist ip range for now).

Those declarations are only allowed at the ingress level so we need to declare multiple
ingresses.

For example, this adaptation currently allows us specify an extra ip range (our vpn) to
access some scheduler endpoints (scheduler metrics, ...) for production and staging
environment (without opening the remaining parts).

This also makes the range ips we declare as a list of ip ranges (instead of csv string).
This allows to comment each range with their definition.

As a temporary workaround for a better filtering per endpoint.

Refs. swh/infra/sysadm-environment#5060