-
Bug
-
Resolution: Unresolved
-
Medium (View bug fix roadmap)
-
None
-
8.6.1, 8.5.3, 8.13.2, 8.20.15
-
8.05
-
15
-
Severity 2 - Major
-
13
-
Issue Summary
At times bulk deletion of projects are carried out to clean up the instance and reduce the load on the index. During deletion of the projects, the instance becomes unusable and causes significant performance degrade. Customer analysis shows that this originates from the large number of queries being run in the database individually rather than in batches and very frequently. The time taken for individual query, even though smaller adds up significantly in total, when the database is located remotely.
Our Analysis shows that this is because to in the delete command, the code around workflows & workflow schemes doesn’t batch the database calls so well. Moreover, we’re often accessing these workflows through caches—yet we clear the caches partway through the process (before dealing with draft workflows) owing to the fix for JRASERVER-8032
Steps to Reproduce
- Create 200 - 300 test projects with one or more test issues using data generator plugin.
- Try deletion of the projects in bulk using a plugin such as REST API or BobSwift CLI, the instance experiences large number of calls to the database contending with other normal user operations.
Expected Results
The project deletion should not cause significant performance degrade.
Actual Results
Project deletion causes the instance to become unusable and causes performance degradation.
Workaround
Currently there is no known workaround for this behavior except to perform the deletion during any available maintenance period if possible.
