Index Corruption

tunderwood89 · **Joined:** Mon Nov 26, 2012 2:34 pm **Posts:** 6

Hi,

We recently started seeing some exceptions pop up and I'm having a hard time understanding what is happening.

We didn't notice this at first but apparently we have been getting a "Unable to Synchronize Source" exception for about the last week then the other morning our software failed to start up getting the following exception.

Code:

java.io.FileNotFoundException: /var/tmp/tool/lucene/index/cnwk.tool.model.submission.Submission/_1g.prx
(No such file or directory)
Caused by: org.hibernate.HibernateException: could not init listeners
Caused by: org.hibernate.search.SearchException: Unable to open Lucene
IndexReader
Caused by: java.io.FileNotFoundException: /var/tmp/tool/lucene/index/cnwk.tool.model.submission.Submission/_1g.prx
(No such file or directory)

Since this caused the software to fail to work in any way we have had to disable hibernate search until we get this sorted out.

The index is being distributed across 3 machines using JMS (Active MQ). We are using Hibernate 3.6.3.Final and Hibernate Search 3.4.0.FINAL (an upgrade to 4.x is in process). All of this is running in Resin 4.0.30.

Any idea what the problem could be and what I can do to fix it? I'll be glad to provide more details if they are needed just let me know!

Thanks!

sanne.grinovero · **Posted:** Mon Apr 22, 2013 12:09 pm

Hi,
are you able to verify if the reference index on the master is also corrupt?
It would help if you could understand if the problem is to be identified in the "copy to slaves" process or if the master produced a corrupt index.

Is there an initial exception which could be the ultimate cause of this? Please try digging in the logs for the first exception.

Besides identifying the problem, are you able to get it to recover? Worse case I'd suggest to delete the indexes and trigger an index rebuild using the MassIndexer on the master.