I'm now seeing that interactive updates are also caught in the sleep-wait-for-maxlag cycle, we need to skip that. That may have been part of the reason for recent slowness all around - if some of the server threads are just sleeping they're not responding to requests!
arthurpsmith
commented
4 years ago
I'm now seeing that interactive updates are also caught in the sleep-wait-for-maxlag cycle, we need to skip that. That may have been part of the reason for recent slowness all around - if some of the server threads are just sleeping they're not responding to requests!