This page covers conditions under which Confluence will become unresponsive and need a restart . It does not cover stack trace error messages thrown from specific uses of Confluence, nor does it cover inability to get Confluence running. Consult Performance Tuning and Garbage Collector Performance Issues if Confluence could be a bit faster or Confluence Startup Problems Troubleshooting if Confluence cannot start.
The first thing to do when Confluence freezes or becomes unresponsive (but the Java process is still running) is to take four thread dumps, separated by 30 seconds between each dump. Thread dumps will show what threads are currently running, (what Confluence is currently doing at the moment the thread dump is taken) and will greatly help diagnosing your issue.
One good tool for analysing a thread dump is TDA.
Known causes for crashes
Running in a virtual operating environment
See our note about virtual environments:
Java Seg Faults
If Java itself can crash, either due to a bug in Confluence or the JDK. See SIGSEGV Segmentation Fault JVM Crash.
Out of Memory: Permgen space
Search the atlassian-confluence.log file for the text
java.lang.OutOfMemoryError: PermGen space. You can solve this problem by increasing your PermGen space to 256M. Make sure to specify PermGen space as opposed to heap space. PermGen space is a straight-forward increase of the memory setting.
Out of Memory: Heap space
Unlike PermGen space, Heap space memory may indicate a memory leak and can require more troubleshooting than just increasing the amount. Managing memory settings requires some analysis; there is no appropriate universal recommendation. To determine if your heap space memory is too low, make sure your XML backup is disabled (see below) and enter plugin safe mode to eliminate memory leak possibilities from our two most likely sources. Once you've done that, read through our suggestions for setting the ideal settings for your environment.
Cluster Panic Messages
Especially in Confluence 3.0.1 or 3.0.2, cluster panics can render the instance inoperable. See 'Database is being updated by an instance which is not part of the current cluster' Error Message.
Plugin related issues
The following plugins have caused suspended threads in some cases. This does not mean these plugins won't work, just that in some cases they may be problematic:
- Team Calendars for Confluence Troubleshooting
- Poor performance due to Confluence Usage Tracking plugin
- Poor performance from excessive garbage collection in the Excel plugin
- User Lister Plugin
- Contributors Plugin
- SQL Plugin
In some cases, these plugins may need upgrading or other maintenance. In some cases they must be disabled. On each plugin's page, you can browse the jira issues. Try entering plugin safe mode for troubleshooting purposes.
Remote API scripts can hammer Confluence into submission! If you've recently added a script, remove it for troubleshooting purposes or disable the Remote API.
Databases: SQL Server 2000 and HSQLDB
SQL Server 2000 has been problematic. Use MSSQL 2005 instead. See our Known Issues for SQL Server.
The in-memory database HSQLDB is for evaluation purposes only. Make sure to migrate to a new database when moving into production.
The XML backup is known to be inefficient. With large instances, it can crash the server. Move to the Production Backup Strategy for an efficient and reliable backup solution.
LDAP and Hibernate cache
Verify that the parameter "cache=true" is configured in your
<confluence-home>/confluence/WEB-INF/classes/atlassian-user.xml file. It should look like this:
This is a hold-over from CONF-9303.
Search Appliance Indexing
Search robots like Google Search Appliance or others can have a heavy impact on Confluence. Disable these as a test, to understand their impact.
Page Restrictions in Large Spaces
If you have page restrictions in spaces with deep hierarchies, you might hit indexing problems when adjusting the page permissions. See Page Restrictions Performance Considerations.
There are a variety of upgrades that have been improved through our versions. Make sure to stay current with Confluence versions. The following bugs are known to have caused crashes or hangs:
- Tangosol Memory Leak - Fixed in 2.8.
- Confluence Hangs While Generating Thumbnail - fixed in 2.9
- RSS Feeds to external websites - fixed in 2.7
- Out of Memory due to large text based attachment - fixed in 2.9
- Page Not Found action too expensive on large installations - fixed in 2.10
-   entries risks blowing the stack - present only in 2.10 and 2.10.1. Fixed in 2.10.2.
- Java quits or exits - Seg Fault due to recursive ExcerptInclude Macro
- Office Connector Freezes Confluence When Loading Excel 2007 File With Date Macro
Upgrade Java Version
Some issues are corrected and others are improved by Java 1.6. Java 1.6 is supported in Confluence 2.7.
Check known performance issues from our Knowledge Base