Description
Issue Summary
Customer is observing high memory utilisation over time. They have been using Jira with max heap of 30 GB with async webhooks and in two or three weeks it is observed that memory utilisation goes to 90% and higher. It leads to back to back Full GC which even are not able to release memory. This overall make Jira unresponsive.
Probably this is due slow webhook endpoints. Customer reports that heap utilisation increases over time. These screenshots are captured from heap-dump when captured in regular intervals(in every 4 day). This is because at time when heap utilisation was more than 90%, we were not able to capture heap-dump and even thread-dump.
Steps to Reproduce
Unfortunately this is not reproducible in test lab or staging environment.
This is reproducible on Data Center: (no)
Expected Results
Heap Utilisation should not be so high that it leads to back to back Full GCs. There should be a configuration to set webhook http async connection SocketTimeout to smaller value so that slow connections can be released. This would garbage collect objects.
Actual Results
High Heap Utilisation.
Workaround
There is no workaround as of now.