Issue in availability metrics in 5k GET GitLab Omnibus installation

Summary

The dashboards of a fresh installed 5k environment does not show information correctly.

Already created in GET: gitlab-environment-toolkit#467 (closed)

Steps to reproduce

Install a 5k environment via the GET.

What is the current bug behavior?

avg_over_time(pg_up{job="postgres",instance=~"$instance"}[$__interval]) or avg_over_time(up{job="postgres",instance=~"$instance"}[$__interval])

overview-actual

postgresql-actual

What is the expected correct behavior?

avg(
  avg_over_time(pg_up{job="postgres",instance=~"$instance"}[$__interval]) or avg_over_time(up{job="postgres",instance=~"$instance"}[$__interval])
)

overview-expected

postgresql-expected

Relevant logs

Details of package version

Will update soon to latest 15.x version. But in 14.9 this is present, maybe also in the newest version.

||/ Name           Version      Architecture Description
+++-==============-============-============-============================================================
un  gitlab-ce      <none>       <none>       (no description available)
ii  gitlab-ee      14.9.4-ee.0  amd64        GitLab Enterprise Edition (including NGINX, Postgres, Redis)

Environment details

  • Operating System: Ubuntu 20.04
  • Installation Target, remove incorrect values:
    • VM: Openstack
  • Installation Type, remove incorrect values:
    • New Installation
  • Multinode (5k reference)

Configuration details