Some index updates get lost when using Oracle Database with Real Application Clusters (RAC):

      enqueue Newly enqueued entry in journal [main_index] has an ID [423954] that should have been higher than the journal state store's most-recent-id [423965]. it is likely that this node's journal state store is corrupt.
      

      The reason for this error is that sequences in Oracle RAC are not sequential by default.

      Workaround

      1. Change SEQ_JOURNAL_ENTRY_ID sequence to be ordered: ALTER SEQUENCE SEQ_JOURNAL_ENTRY_ID ORDER;
      2. Rebuild indexes.

            [CONFSERVER-35764] Index updates can get lost on Oracle RAC

            Hi mvdkleijn,

            I apologise for the slow answer. This fix only works for installations that have not been upgraded to 5.6. The resolution 2 in Index not working due to 'journal state store is corrupt' error KB page should fix the issue for installations that have been upgraded to 5.6.

            Olli Nevalainen added a comment - Hi mvdkleijn , I apologise for the slow answer. This fix only works for installations that have not been upgraded to 5.6. The resolution 2 in Index not working due to 'journal state store is corrupt' error KB page should fix the issue for installations that have been upgraded to 5.6.

            This appears to still be a problem in Confluence 5.7.4.

            2015-06-30 15:54:34,646 WARN [http-bio-8090-exec-1062] [confluence.impl.journal.DefaultJournalManager] enqueue Newly enqueued entry in journal [main_index] has an ID [683655] that should have been higher than the journal state store's most-recent-id [683709]. it is likely that this node's journal state store is corrupt.
            

            Would this be fixed for upgraded systems as well or only for completely new installs?

            Deleted Account (Inactive) added a comment - This appears to still be a problem in Confluence 5.7.4. 2015-06-30 15:54:34,646 WARN [http-bio-8090-exec-1062] [confluence.impl.journal.DefaultJournalManager] enqueue Newly enqueued entry in journal [main_index] has an ID [683655] that should have been higher than the journal state store 's most-recent-id [683709]. it is likely that this node' s journal state store is corrupt. Would this be fixed for upgraded systems as well or only for completely new installs?

              drizzuto David Rizzuto
              onevalainen Olli Nevalainen
              Affected customers:
              0 This affects my team
              Watchers:
              7 Start watching this issue

                Created:
                Updated:
                Resolved: