Tar MK Corruption

[jakej40603950]

As given in all the documentation, the solution to TarMK corruption is to run the 'check' run-mode on oak-run jar and revert the repository to the last known good revision by making changes to journal.log.

But doing this would wipe out some of the recent data from the repository. Is there a way to know what data or content has been erased by reverting to an earlier revision? Also what would be the best way to get this data back?

Thanks.

[Jaideep_Brar]

There is another way to find the corrupted nodes and remove them manually. Check the "B. Remove corrupted nodes manually." at [1]

[1] Offline Compaction fails with SegmentNotFoundException & IllegalArgumentException

[berliant · Accepted solution]

When you determine the latest good revision, you can find the approximate date and time of that revision:

Offline Compaction fails with SegmentNotFoundException & IllegalArgumentException step 4.

Next, you can query your JCR by date of creation and modification. For example, for cq:Pages you can use a query similar to:

/jcr:root/content//element(*,cq:Page)[(jcr:content/@cq:lastModified >= xs:dateTime('2015-11-03T04:00:00.000-05:00') and

jcr:content/@cq:lastModified <= xs:dateTime('2015-11-03T20:00:00.000-05:00')) or (jcr:content/@jcr:created >= xs:dateTime('2015-11-03T04:00:00.000-05:00')

and jcr:content/@jcr:created <= xs:dateTime('2015-11-03T20:00:00.000-05:00')) ]

The result can be packaged via Query Packager

That is the best option to backup the content that will be wiped up by reverting to the latest good revision.

[jakej40603950]

This helps a lot thanks.

[jakej40603950]

Thanks. In this case (approach B) how do you find which data is lost and what'd be the best way to restore that?

[Jaideep_Brar]

When you run the groovy script, It will list the paths that have been corrupted.

You can then delete those paths using the groovy script and create a package of these nodes from other instances.