You are viewing a plain text version of this content. The canonical link for it is here.

Posted to oak-issues@jackrabbit.apache.org by "Michael Marth (JIRA)" <ji...@apache.org> on 2015/04/20 13:35:59 UTC

[jira] [Created] (OAK-2788) Improve system resilience in case of index corruption

Michael Marth created OAK-2788:
----------------------------------

             Summary: Improve system resilience in case of index corruption
                 Key: OAK-2788
                 URL: https://issues.apache.org/jira/browse/OAK-2788
             Project: Jackrabbit Oak
          Issue Type: Improvement
          Components: lucene, query
            Reporter: Michael Marth


I have had issues in cases when the async Lucene index had gotten corrupted. With this index unusable the only option was to re-index.  The problem however was that there were ongoing queries relying on this index even during re-indexing. Because the original index was corrupted these queries led to further load on the system (traversals afair).

I wonder if we could improve the system resilience in such situations.
One thing I could think of: could we maybe fallback to the last known non-corrupted index state while the re-index is running? This would at least take off the load due to new incoming queries.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)