-
Bug
-
Resolution: Unresolved
-
Medium
-
None
-
7.19.0, 8.5.0, 8.6.0, 8.8.0, 8.7.0, 8.9.0
-
2
-
Severity 2 - Major
-
Issue Summary
Flush Edge Index Queue Job transaction runs for several days and so edge index processing is halted.
This is reproducible on Data Center: (yes)
Steps to Reproduce
- Create few large page trees under Confluence(Eg 5 large page trees with 10K child pages each)
- Change the permissions across these parent pages and toggle these few times. Try out 10 different iterations.
Expected Results
Flush Edge Index Queue Job queues these changes and processes them in batches one by one.
Actual Results
Flush Edge Index Queue Job pushes these changes to the same session time and again causing the entities managed by hibernate session to grow. If the sessions entities grew really large in small amount of time, this can cause recursive behaviour(hibernate dirty check mechanism which would walk through each entity) where the transaction would never complete and so this directly impacts edge index processing under Confluence. In this specific case, we could see hibernate was trying to work with 1,090,303 entityEntries as a part of flushEntities method based on heap dumps.
Workaround
Copy indexes from the working node manually.
| Form Name | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|