Martin Braun

@DimCitus while this is probably not on the roadmap for some time, if someone were to try to do this on their own, what steps would be needed to take...

I 100% understand your points and think this is a good route to take for the space that pg_auto_failover is trying to take. It's simple and predictable. I actually think...

Something that comes to mind from a marketing standpoint though is that it would be easier to "sell" the fact that the monitor is not HA if there was a...

Just something that might be of interest for people that are concerned with this issue: A simple way to get alerted on these issues where you might have to manually...

@DimCitus this would be a service that is included in the pg_autoctl binary (=> so it would be required to be written in C) or could this be a separate...

This article https://luppeng.wordpress.com/2020/11/11/building-resiliency-in-the-pg_auto_failover-monitor/ raises an interesting point which I think could be explored. While I don't think everyone wants to run their postgres application inside Docker and/or Kubernetes because it...

If pgaf does not support connecting to multi host connection string monitors, one can always rely on HAProxy to select any viable IP for the monitor. I feel like this...

Which version of libpq are you using? I recently did the same with our django app and needed to upgrade libpq for the HA behaviour to work properly.

+1 on separating the autoscaler: I think this would be a great standalone project which would help the Docker Swarm ecosystem in general.

Thanks for the link. Your script looks good, but there is something about such things written in bash that makes me uneasy. I guess it is fine, but a solution...