V2 metrics cluster dashboard reports the number of shards up as the database count

[collinfunk]

Using a cluster with one database with two shards, the V2 metrics cluster dashboard mistakenly reports that I have two databases:

This is because it uses redis_server_up, which is the number of shards that are running, documented here.

I have an idea for a fix, will submit a PR.

[collinfunk]

My idea was to use cont(db_config{cluster="$cluster"}), but that won't work since it doesn't change the status if a database is unavailable.

[slorello89]

@collinfunk - which dashboard is this exactly? You should be able to do something like a min by(cluster,db)(redis_server_up) < count by(cluster)(db_config) to get approximately what you're looking for if an (e.g. whether there's any unhealthy redis databases)

[collinfunk]

@slorello89 The one that I screenshotted in the original report is here.

The V1 dashboard here uses count(bdb_up{cluster=\"$cluster\"}), which gives the behavior that I expected from the title:

I assumed they were meant to be the same, but maybe it was intentional. I'll see if I can try your suggestion. Thanks!