[JCR-3905] Journal sync after full re-index does not save journal revision

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.8.1
• Component/s: clustering, indexing, jackrabbit-core
• Environment: SQL persistence store
• Assignee: Unassigned
• Reporter: Philip Wasson
• Created: 31/Aug/15 21:39
• Updated: 02/Sep/16 14:12

Description

In org.apache.jackrabbit.core.query.lucene.SearchIndex.doInit, if it discovers that the index is empty, it creates an initial index and then calls checkPendingJournalChanges(). That method reads all the new journal records (all at once, which is separate bug JCR-2483) and applies them. Since it is now caught up, it should save the revision number that it is up to, but it appears not to. As far as I can tell, this means that when the next journal sync occurs, it will re-fetch and re-process all the records it just processed.
Also, it's not clear why it bothers to process the journal at this stage anyway, instead of letting it happen at the next journal sync.

Issue Links

• is related to: JCR-3162 Index update overhead on cluster slave due to JCR-905
• is related to: JCR-2483 Out of memory error while adding a new host due to large number of revisions