Apologies. Yes, I meant the corresponding worker process. I neglected to mention that it also seems like distribution of work to the various workers does not take into account the busy-ness of any particular worker, so hypothetically more work could be directed to this already bogged down process.
For the record we spotted an issue with 3.4.x Bug #2070304 “regiond at 100% CPU after UI reconnect causing API...” : Bugs : MAAS . With high probability the issue you have faced was related to it
1 Like