Use upstream v0.8.0

[benbz]

Our fork changed how pg_current_wal_lsn is generated, whereas prometheus-community/postgres_exporter/pull/307 exposed the same value as pg_current_wal_lsn_bytes. So we'll need to update any dashboards / alerts

Also remove the need for our fork by pulling matrix-org/postgres_exporter@f77ce19 into queries.yaml

[michaelkaye]

This looks OK to me; worth trying on the smaller hosts for a bit, especially lethe which has none to check it's OK, before deploying to the rest.

Additionally we should avoid deploying this on friday to prevent it interacting with our postgresql cache size changes - we need clean before/after knowledge for that to be monitored safely.

[benbz]

Additionally we should avoid deploying this on friday to prevent it interacting with our postgresql cache size changes - we need clean before/after knowledge for that to be monitored safely.

I'm going to avoid merging & deploying this at all prior to Friday - given we're going to be running postgres playbooks and it could be difficult to skip this change if it is merged and checked out on ansible box

[michaelkaye]

Also before deployment, NB we inlined this.