Issue Summary

This is reproducible on Data Center: (yes)

We noticed that when we inject several transactions in a row of one automation that creates issues and links the trigger issue and the new one with a high heap and CPU utilization, causing a severe performance issue.

Steps to Reproduce

1. Create a new automation as per the example below
   
2. Execute a POST to inject comments on issues to force the automation to run

   curl -D- -u MY_USER:MY_PASS -X POST --data "{\"body\": \"Comment added when resolving issue\"}" -H "Content-Type: application/json" http://localhost:49414/j9414/rest/api/2/issue/DUN-23/comment
   

 Note: We created this simple scenario as a test, but in the field, we see the same issue when customers are using the trigger "Schedule based on JQL" hence a lot of issues will be injected transactions as described above and cause the same behavior.

Here is the heap dump as an example of where this type of flow will cause an OOM error:

Expected Results

Automation to run and not cause huge heap and CPU utilization, basically we need performance improvements.

Actual Results

High GC activity followed by OOM errors and huge CPU utilization

Workaround

Currently, there is no known workaround for this behavior. A workaround will be added here when available