Performance Problems due to Cluster Communication
This article applies to Confluence clustered 5.4 or earlier.
Symptoms
When taking a thread dump, some threads look like:
Thread: TP-Processor40 state=object_wait
java.lang.Object.wait
java.lang.Object.wait
com.tangosol.net.cache.OverflowMap$Status.waitForAvailable
com.tangosol.net.cache.OverflowMap.prepareStatus
com.tangosol.net.cache.OverflowMap.beginKeyProcess
com.tangosol.net.cache.OverflowMap.getInternal
com.tangosol.net.cache.OverflowMap.get
com.tangosol.coherence.component.util.CacheHandler.getLease
com.tangosol.coherence.component.util.CacheHandler.getCachedResource
com.tangosol.coherence.component.util.CacheHandler.get
com.tangosol.coherence.component.util.SafeNamedCache.get
com.atlassian.confluence.cache.tangosol.TangosolCache.get
Cause
Nodes in the cluster are not communicating.
Resolution
See Data Center Troubleshooting. If you are on a version prior to Confluence 3, consider an upgrade. Performance has been greatly improved in that version.
Last modified on Mar 30, 2016
Powered by Confluence and Scroll Viewport.