The function log_append_buffer2 has been shown to cause large latency spikes when flushing, which occurs "in the path" of searches and results being sent.This can cause operation latency spikes and delays.
This behaviour should be improved.
From response time, results are logged after the results are sent back. So it should not create contention for end of operations but it will contribute to contention on starting operation. Is the latency you observed occurs because the effective processing of an new operation is delayed until its start is logged ?
Metadata Update from @tbordaz: - Custom field origin adjusted to None - Custom field reviewstatus adjusted to None
There are access log's writen an the start of the operation that can delay a client, and it also causes a delay in the time for a worker to be returned to the work pool (and other workers to catch up).
This in mind, the connection freelist seems like the more important change to make.
Metadata Update from @mreynolds: - Issue set to the milestone: 1.4 backlog
389-ds-base is moving from Pagure to Github. This means that new issues and pull requests will be accepted only in 389-ds-base's github repository.
This issue has been cloned to Github and is available here: - https://github.com/389ds/389-ds-base/issues/3840
If you want to receive further updates on the issue, please navigate to the github issue and click on subscribe button.
subscribe
Thank you for understanding. We apologize for all inconvenience.
Metadata Update from @spichugi: - Issue close_status updated to: wontfix - Issue status updated to: Closed (was: Open)