Open claudijd opened 7 years ago
I'd prefer to have both thread and process control defined in the worker config so we can scan at both the thread and process level.
We also need full-stop way of trapping exceptions on workers so they never get dead-ended on an unexpected stack trace and are restarted. This is effectively a bare rescue of sorts, but we want a way to report this somewhere so the information isn't lost and we can see the highest hitting bugs to go fix.
The workers are not as robust as I would prefer. Things that I think we could do better include...