The mixin dashboard uses metrics that don't exist

[snarlysodboxer]

What did you do? Run 0.13.2 in Kubernetes as a sidecar container via the postgres-operator. Load this dashboard: https://github.com/prometheus-community/postgres_exporter/blob/master/postgres_mixin/dashboards/postgres-overview.json

What did you expect to see? Something I could further tweak.

What did you see instead? Under which circumstances? It appears all or nearly all the metrics this dashboard is looking for don't exist or are not being exported. - There are potentially similar ones being exported, but I might as well be starting from scratch because I can't know for sure if for example pg_stat_database_conflicts_confl_deadlock should replace pg_stat_database_deadlocks or what.

Environment

• System information:

  Linux 5.10.104-linuxkit x86_64

• postgres_exporter version:

  quay.io/prometheuscommunity/postgres-exporter:v0.13.2

• postgres_exporter flags:

--log.level=warn
--log.format=json

• PostgreSQL version:

  12,13,14

• Logs: no error logs

[sysadmind]

The dashboard is something I haven't worked with much, but I did attempt to get it loaded locally. Running a local postgres V15, the master branch of the exporter, and the latest tag for both prometheus and grafana, I was able to load the dashboard. The only panel that was not working for me was the buffers. All of the other metrics were there for me. I'm not sure where pg_stat_database_conflicts_confl_deadlock comes from. You mentioned it, however the dashboard that you linked to doesn't reference it at all.

I did open #911 to hopefully make that dashboard easier to work with moving forward.