PG multi region cluster log spam


We’ve created a PG cluster in 3 regions (2x CDG, 1x ORD, 1x SIN) following the documentation.

Everything is working fine except for one small issue, we’re being spammed in the logs of the leader instance (in CDG) by the following warning.

app[<ID>] cdg [info] sentinel | 2022-09-07T05:46:34.101Z WARN cmd/sentinel.go:276 no keeper info available {"db": "<ID>", "keeper": "<ID>"}

This line appears every 10 seconds.

We have no idea of what might be causing this, we’ve tried to remove and create the cluster again, but after a deployment/restart of the leader, the warning comes back.

We have not seen yet any issue with the database in itself but this is a new application we’re currently developing on Fly to test it out.

Since we are using OpenTelemetry for logs, metrics and traces and we ship all of those to 3rd party services, we want to minimize noise (and cost) by having logs as clean as possible. So we’d love to understand this warning and fix it.