Related to swh/infra/sysadm-environment#5613

We will use the old cnames for the new services and we may want to compare the
previous versions with the current.

Refs. swh/infra/sysadm-environment#5608

Could help diagnose swh/infra/sysadm-environment#5570

Related to swh/infra/sysadm-environment#5606

Related to swh/infra/sysadm-environment#5606

That one was already functional since it did not rely on the lookup. This just
aligns its definition with the other two.

Refs. swh/infra/sysadm-environment#5588

All checks were deployed on the production environment because it relied on
lookup of variables changing between environments. As the checks are deployed
on pergamon which is a production node, that could not be functional.

Refs. swh/infra/sysadm-environment#5588

Given the metrics in prometheus, the load of maxxi is
very high when a graph compression is in progress.

It generate a lot of useless alerts we never check.

The ELK disk limit is configured to 10%.
The shard redispatch is async so it's possible the disk occupation
could be temporary more than 10%.
There is no need to trigger any alert when it's the case

Related to swh/infra/sysadm-environment#5585

Since the migration to bookworm, it comes with a node 18 version which is incompatible
with the yarn coming from bookworm so pinning to a most recent version still compatible.

Refs. swh/infra/sysadm-environment#5415

Some origins can take several days to be ingested (svn, ...) so the default
30mn timeout is way to small.

Related to swh/infra/sysadm-environment#5595

Related to swh/infra/sysadm-environment#5426

Related to swh/infra/sysadm-environment#5426

Refs. swh/infra/sysadm-environment#5426

Close swh/infra/sysadm-environment#5587

Without the typo, this is actually ok to not use the full path.

Refs. swh/infra/sysadm-environment#5547

Refs. swh/infra/sysadm-environment#5547

Refs. swh/infra/sysadm-environment#5547