Wikimetrics does not seem to recycle either celery workers or mysql connections properly. This seems like a new problem since things seemed fine a short time ago. But that could also be because we weren't getting as much traffic as we are now.
Version: unspecified
Severity: normal
bingle-admin wrote:
Prioritization and scheduling of this bug is tracked on Mingle card https://wikimedia.mingle.thoughtworks.com/projects/analytics/cards/1326
The connection starving problem turned out to be due to a labs db configuration revert. This is fixed now. Closing connections is also happening more reliably all over the wikimetrics code base, but I believe this was not part of the root cause here.