I can't seem to pinpoint this one. Occasionally some jobs fail with an End Marker Unreadable error and when they do they fail at the exact same byte count (with they exception of two occurances, but even they matched each other). The interval seems so random...I've seen it fail twice in one day once, but beyond that it happens at least once every 2-4 weeks from the last time, but at no pattern. The best part is that I can watch a job run, let it fail, and then restart the job without problem.
I've ruled out the tapes because all my tapes have been ending up EMU, and most of them are already being written to by the exact job the day before. The tape unit's fine because, obviously, it's writing data to all the tapes at some point. There are two exact jobs (one writes to rotation media and the other writes to offsite media) that run from a single policy, but each job uses the same selection list. The selection list is only for one server, which is also the BE 10.0 host.
The error report gives V-79-57344-34028 for a cyclic redundancy check. But I find it impossible that any physical, electrical, or hell, even theoretical or godly event could cause the same problem at the same point...unless Symantec sold its soul to the devil (possible) and BE is reporting bad information on the byte count. The only thing that the tapes might have in common is that they seem to be EMU-ing at around tape capacity, but even that might be an inaccurate statement as I've seem some fail after only a few gigs of data.
Any ideas?
