Corrupt Tar Journal File

[richard_gray]

We have a problem with a corrupt file in the tarJournal of our two-node author cluster (CQ5.4, CRX2.2.0.70), and I was hoping someone might have some advice on how to recover from this situation. Our nightly tar optimisation run gets stuck on the corrupted file, which is causing the tarJournal to grow and grow. The CRX log is filled with messages like this:

17.01.2014 03:10:24 *ERROR* TarFile: Truncating tar file /opt/app/cms/author/crx-quickstart/repository/tarJournal/data_00714.tar from 254326272 to 254325248 (TarFile.java, line 374) 17.01.2014 03:10:24 *WARN * TarFile: Error reading entry name /opt/app/cms/author/crx-quickstart/repository/tarJournal/data_00714.tar id:714 length:254326272 append:254325248 3096610 03 (TarFile.java, line 371) java.io.IOException: Invalid entry name: -_20I????^@^@^Rslingevent:created^@^@^@^@^@^@^Nslingevent:Job^@^@^@^@^@^@^Eadmin^@E^DI^@^@^@^A^@^@~/var/eventing/jobs/anon/3c in /opt/app/cms /author/crx-quickstart/repository/tarJournal/data_00714.tar pos: 254325760 at com.day.crx.persistence.tar.file.TarFile.readEntryName(TarFile.java:253) at com.day.crx.persistence.tar.file.TarFile.scanIndex(TarFile.java:297) at com.day.crx.persistence.tar.Optimize.optimizeBlocks(Optimize.java:107) at com.day.crx.persistence.tar.OptimizeThread.loop(OptimizeThread.java:463) at com.day.crx.persistence.tar.OptimizeThread.run(OptimizeThread.java:382) at java.lang.Thread.run(Thread.java:662)

Does anyone have any recommendations on how to fix this problem? Would either of the following options work?

1. Delete the corrupted file and then manually run tar optimisation.
2. Delete the contents of crx-quickstart/repository/tarJournal on the master and then reinitialise the slave from the master.

Aside from the growing journal, the cluster appears to be functioning normally. Any help would be appreciated.

Thanks,
Richard

[smacdonald2008]

Here is a KB that may help you:

http://helpx.adobe.com/experience-manager/kb/performancetuningtips.html#TIP09

[richard_gray]

Thanks for the reply.

I'm not sure the tips there are useful though, as we are using TarJournal rather than FileJournal. Also, the old files in our tarJournal are now older than our (default 1M) maximumAge setting. I believe the corrupted file is preventing that cleanup from working. Any thoughts on how I can manually clean the journal and start afresh?

Would deleting the tarJournal and then re-establishing the cluster work?

[smacdonald2008 · Accepted solution]

See if this older community thread is useful:

http://forums.adobe.com/message/4599467