Uploaded image for project: 'Bitbucket Data Center'
  1. Bitbucket Data Center
  2. BSERV-7420

Improving handling of OOME in certain contexts

XMLWordPrintable

      When an OutOfMemoryError occurs in certain contexts, the Hazelcast instance is terminated which prevents Stash on that cluster node from potentially recovering from the error.

      OutOfMemoryErrors are not always recoverable anyway, but this behaviour of Hazelcast when it is caught in these contexts should be made more robust. This will remove some occurrences of the following log error:

      2015-01-05 01:36:21,091 ERROR [threadpool:thread-36072]  c.a.s.i.c.StateTransferringExecutor Error while processing asynchronous task
com.atlassian.cache.CacheException: com.hazelcast.core.HazelcastInstanceNotActiveException: Hazelcast instance is not active!
        at com.atlassian.cache.hazelcast.HazelcastCachedReference.get(HazelcastCachedReference.java:95) ~[HazelcastCachedReference.class:na]
        at com.atlassian.stash.internal.license.LicenseServiceImpl.get(LicenseServiceImpl.java:101) ~[LicenseServiceImpl.class:na]

              cszmajda Cristan Szmajda (Inactive)
              drohan Daniel R
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

                Created:
                Updated:
                Resolved: