In regard to: Re: [Networker] Tapes Marked Full Prematurely when Cloning,...:

Going back through the logs, it does appear to only be /dev/nst1 each
time.

We're having an exceedingly strange issue with /dev/nst1 on our primary
backup server too, and I only discovered it by chance.  Processing
the daemon.raw file in the way I mentioned is what gave me the evidence
I needed to go to our vendor about the issue.

  I'll touch base with HP again and see if I can't get the drive
swapped.

Yeah, good idea.  Like I said earlier, if you haven't already done so,
try downloading "Device Data" using HP's LTT.  Something in there might
stand out.  There's lots of useful information.

   Yes it is only on the clones, which leads me to believe that
tape to tape cloning puts more of a demand on drives which may show an
error that it would otherwise compensate for.

Depending on your network, the drives could very well be able to stream
at a higher rate when cloning drive-to-drive.  If both drives are on the
same bus it might also be placing much higher stress on the data path
than a typical backup would.

Tim

Tim Mooney <Tim.Mooney AT NDSU DOT EDU> Friday, June 14, 2013 4:26 PM >>>

In regard to: [Networker] Tapes Marked Full Prematurely when Cloning, Eric...:

We have an HP MSL4048 Jukebox with 2 drives.   Up until earlier this
year the backups fit two tapes and then they were cloned and life was
good.    Since we have gone to 3 tapes, we are having more and more
clone failures due to "tape marked full prematurely".   I've gone down
the support road with HP and did all the firmware updates, tried new
tapes, etc. to no avail.

I assume HP had you use the "Library Tape Tools" (LTT) to run drive
diagnostics on the tape drives and those passed.

Did they also have you download the device data for the tape drives?  It
essentially uses SCSI commands to query the drive and get a wealth of
information about drive health, lifetime mount count, etc.  If you haven't
already done that, reviewing that information may prove useful.

Also, is it both tape drives that are marking tapes full prematurely, or
is it only one?  You can tell if you process your daemon.raw file with
a script.  You want to look for the text patterns

        'on %s is full'
        'used %s of %s capacity'

those are the two lines that will allow you to figure out which drive
was writing to the tape when it filled.  You might discover that it's
just one of the two tape drives that's causing all the problems.

Here's the odd part (to me), I do a backup on friday and when the friday
clone fails, I can take those friday clone tapes and label them for
monday as the backup tapes and they hold the full amount.    Has anyone
experienced anything like this?

So really, the problem only manifests during cloning, not when doing
backups?

Tim

--
Tim Mooney                                             Tim.Mooney AT ndsu DOT 
edu
Enterprise Computing & Infrastructure                  701-231-1076 (Voice)
Room 242-J6, IACC Building                             701-231-8541 (Fax)
North Dakota State University, Fargo, ND 58105-5164