Cluster Panic due to Performance Problems
There are multiple causes for this issue. Start with the parent article Confluence will not start due to fatal error in Confluence cluster.
Confluence throws the following error on a single-node deployment spontaneously during production usage:
Fatal error in Confluence cluster: Database is being updated by an instance which is not part of the current cluster. You should check network connections between cluster nodes, especially multicast traffic.
There are several performance problems that can cause this error. They are listed in the order of most likely to least likely:
- If you are on Confluence 3.0.1 or 3.0.2, you are most likely experiencing a bug. See Cluster panics (Non Clustered Confluence 2.10.4, 3.0.1 and 3.0.2).
- The database is overloaded or the database connection has been severed.
- An out of memory error has occurred.
- In some conditions, severely poor garbage collection performance can contribute to this.
- The daily sql dump backup job (in a native tool) is running. This is very unlikely, and should be addressed only as a last resort.
Run Confluence as a single node cluster.
There are multiple resolutions for this problem:
- Ensure you have the patch from Cluster panics (Non Clustered Confluence 2.10.4, 3.0.1 and 3.0.2).
- If you have the word 'Deadlock' in your atlassian-confluence.log, see Troubleshooting deadlocks in Confluence.
- Upgrading or optimizing the database has corrected this in many situations.
- If you find OutOfMemory in your atlassian-confluence.log, see Fix Out of Memory errors by Increasing Available Memory.
- Assess poor garbage collection performance. Check Garbage Collector Performance Issues.
- Only in rare circumstances, and when you can reliably replicate the problem, the workaround is to take the application offline while generating the backup.