Comment 5 for bug 1702671

Also, logs provided in #1 seem to be updates from Maas due to being unable to make full connectivity between controllers. There is one log that stands out specifically and that is this:

ERROR: requested WAL segment 000000010000002100000042 has already been removed

That to me indicate Postgres replication issues, and those replication issue could actually be the reason of why you have cpu starvation. I would like to see full logs but otherwise this doesn’t look related to the originally reported issue.