When a db exception is thrown while full index rebuild runs, the indexing completes, but no further incremental indexing jobs are executed.

It seems that the exception is not handled well, which results in the problems with future indexing jobs.

The issue is discussed at http://forums.atlassian.com/thread.jspa?threadID=40706&tstart=0&messageID=257331096

Similar issue were also observed when our server ran our of local disk space on the mount where the index is stored.