The fix for this bug will be backported to our current Long Term Support releases if it passes our risk assessment.
We recognise the impact of this bug and plan to backport the fix to our Long Term Support releases. Once the exact fix version is known we will update this ticket. Not all code changes are suitable to be backported and this may still be rejected during our review process.
Issue Summary
Create Space request can cause performance issues if there are large amount of indexing queue lag.
This is reproducible on Data Center: Yes
Steps to Reproduce
- Setup an environment and try to push a lot of content(create new pages, comments etc) using REST API script.
- Ensure content indexing queue is lagging by some good amount(20k or more).
- Initiate a create space request.
- It will take a long time till entire indexing queue is processed.
- In the mean time, try to push some more create space request and you would observe all the threads are stuck as initial create space request is holding the index lock.
- Apart from above, if we are using Questions for Confluence plugin, we will see lots of question plugins request also waiting for the lock too. This is similar to issue as raised under CONFSERVER-66182 but instead of reindex thread, the lock is now being held by the create space request under this scenario.
Expected Results
If there is an issue with one space create request, then that should not cause other space create request to halt. In addition to it, this request should not effect the requests from other plugin modules.
Actual Results
One create space request can cause other space create requests or questions plugin request to halt processing if it holds the reindex lock for a long time.
Logging details
We can see below log snippet from atlassian-confluence.log that further confirms that the create space request does not release the index lock for almost 51 hours and it is still further trying to process other indexing items:
2023-07-14 14:22:57,978 WARN [https-jsse-nio2-8443-exec-92] [search.lucene.queue.JournalIndexTaskQueue] flushQueue Flushed 169000 tasks out of 279893 (186642926 ms over lock timeout) -- url: /confluence/rest/create-dialog/1.0/space-blueprint/create-space | referer: https://XXXXXX/confluence/display/SPACEKEY/PAGE-NAME-XX | traceId: c6a5461caf5907d6 | userName: XXXXXXXX
Workaround
Currently there is no known workaround for this behavior. A workaround will be added here when available
- relates to
-
CONFSERVER-97286 Attempting to create a Personal space while incremental/background indexing is already flushing/committing data can lead to a slowdown in indexing rates
-
- Closed
-
- mentioned in
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...