Nikita Kazmin
Nikita Kazmin
From what I remember, failover isn't instantaneous, so the issue you're probably encountering is: 1. Primary is killed, repmgr begins to make sure standby is up to date 2. While...
> @vonzshik you're saying we correctly update the primary state to offline during Break() but we don't consider the next host in the list a primary yet so it fails...
The only question I have is, do we break-break, or cancel-break?
> BTW in theory we could consider trying to send a PG cancellation - but without waiting for it. I'm not so sure it's important that we do that (since...
> Hello, is there any missing work to be done so that this PR is merged? Hello, don't think so, I'm just busy with personal stuff for a bit. Will...
Hello. At first thought seems reasonable, I'll put this on 8.0 and discuss with the team.
It seems like this was fixed by #4964 so we can probably close it?
One way to suppress them is to write a custom processor for OpenTelemetry. The one I've made does use reflection (because `DiagLinkedList` is an internal class) but all in all...
@Brar just in case, did you intentionally close the pr?
Was fixed at some point, closing.