Jump to content

VXA3 tapes, 206 dirty heads, and erased media which isn't


Recommended Posts

Hi,

 

I've had ongoing problems with X23 tapes in a vxa 3 drive with autoloader on an 10.4.6 XServe with UL4D. UL4d firmware has been updated, so has the autoloader and drive. Retrospect is server version 6.1.126. RDU is v 6.1.5.102.

 

The vxa 3 was new in February. It worked ok for a couple of months, but now often indiates tapes with data as erased or has trouble writing. We have lost data on a number of different brand new tapes because retrospect shows them as 'erased' after showing a '206, dirtyheads' warning We haven't had any good tape backups for several weeks.

 

I've been working with exabyte tech support. Some problems were caused through dirty heads on the drive. All our tapes have been brand new in the last 3 months or so (the drive was new in Feb). So, I've been running a lot of drive cleaning tapes, and the drive runs ok for a few days, but then the tape becomes unreadable again.

 

The most recent thing exabyte support had me do (end of last week) was relocate the autoloader in the rack so it was away from strong heat sources, ie. our Xserves.

 

Then on Friday Retrospect hung doing a network backup. It hung pretty hard on 'closing' from a client Mac. I had to force-quit Retrospect, then restart the Xserve to get Retrospect to see the drive. On restarting Retrospect and trying to continue backup to the same set I got messages that the set was out of sync and needed rebuilding.

 

After almost 30hrs of rebuilding on a single tape just 15453 files, 3.8Gb were rebuilt.

 

The rebuild went very quickly at first but didn't do anything for about 26 hrs. The tape was rewinding and locating in the drive but the file count was not going up. The autoloader showed the tape as 'locating', etc.

 

I eventualy stopped the rebuilding process and found the following in the log:

 

 

+ Executing Recatalog at 05/16/2006 12:27

To backup set 2006.a1…

Trouble reading: “3-2006.a1” (197348), error 206 (drive reported a failure: dirty heads, bad media, etc.).

Trouble reading: “3-2006.a1” (197348), error 206 (drive reported a failure: dirty heads, bad media, etc.).

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

Trouble positioning: “3-2006.a1” (227895), error 212 (media erased).

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

...MANY, MANY MORE JUST LIKE THESE...

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

Trouble reading: “3-2006.a1” (227895), error 212 (media erased).

Additional error information for device "Exabyte VXA DC" [0:2],

Sense > f0 00 03 00 00 00 01 24 00 00 00 00 00 05 00 00 00 00

(EXABYTE |VXA-3 |3200)

05/17/2006 16:13:00: Execution stopped by operator.

Completed: 15455 files, 3.8 GB

Performance: 3.9 MB/minute

Duration: 27:45:32 (11:42:43 idle/loading/preparing)

 

 

I can see how Retro may have contributed to the initial network hang problem. But the subsequent drive problem on rebuilding looks more like a hardware problem than a software problem to me. Wheer do you think I should be looking to fix this problem?

 

Thanks,

 

techfish.

Link to comment
Share on other sites

Many thanks for your reply.

Hmm... the drive is set to 2 and the autoloader is set to 11.

I contacted exabyte after reading your post to check what they recommend. They say 2 for the libary and 3 for the drive.

I will try libray in 2 and drive in 3 and let you know.

Is there a particular reason wht you are using drive in 2 and library in 5?

Thanks for pointing me in a new direction,

tf.

Link to comment
Share on other sites

I changed the scsi id numbers and ran the cleaning tape.

 

I decided to rebuild the set due to the out-of-sync error (see my OP) and unrecognised tape, and went back to the original first tape. This tape had never had any problems. But when I inserted it Retrospect showed it as unrecognised.

 

Sigh...

 

I sent more logs off to exabyte after this failure. This really looks like a hardware issue I guess.

 

Regards,

 

tf.

Link to comment
Share on other sites

  • 3 weeks later...

I think that there may be two issues here. You don't give firmware versions for your drive and autoloader, so I don't know whether you are up-to-date on them. Our drive and autoloader firmware was not up-to-date when we received ours. We've got a slightly different setup, but I'll give you our experiences (admittedly, the VXA-2 and VXA-320 drives are on different firmware branches):

 

Xserve G5 2.0 GHz single processor 2 GB RAM, 10.4.6 Server

ATTO UL4D

VXA-2 1x10u PacketLoader Plus with SCSI interface

Library firmware: V1A10D (2/06/06)

drive firmware: 210D (1/10/06)

Retrospect Workgroup 6.1.126, RDU 6.1.5.102

 

We saw numerous dirty head error messages back in December 2005 with that drive firmware update, and those disappeared with the January drive firmware update. I think that some of the error recovery routines were improved. Like you, we clean our drive regularly and use new tapes, replacing tapes each month when they fill, never reusing tapes. The problem persisted across various Retrospect updates, and I think that the fix was done by Exabyte, not Retrospect.

 

The rewinding / media erased error is something different, and we have seen that, too. Reasonable minds may differ as to whether it's a Retrospect problem or an Exabyte problem, but we see that error if there is a tape in the drive when Retrospect starts, if the tape is not at BOT. For that reason, we have Retrospect set to rewind on quit and to eject the tape back to the autoloader. It's still possible to see that error if, for example, there is a power failure during a backup that causes the server to shut itself down (even if done gracefully while on UPS), leaving the tape not at BOT, which is how we saw it the first time. I've put in a feature request for a future version of Retrospect to have a preference to do a "rewind on start" before looking at the tape as a workaround, to ensure that the tape is always at BOT when Retrospect goes out to do its initial device scan (in other words, to repeat the device scan after an iniitlal rewind). But somehow the Exabyte drive's status bits report back to Retrospect that the media is erased when starting away from BOT, I suspect because of some confusion because some expected BOT header information isn't there to allow the drive to tell what type of tape is in there (with the recent firmware updates, the drive looks at header information to determine the VXA-1, 2, 3 type of the tape and other info such as whether the tape has been marked as WORM). When in this mode, the drive (or Retrospect, I don't know what the exact source of the error is) will cruise past logical EOT onto blank tape, and keep trying to find something, with tape motion in both directions during this period of confusion. Only cure seems to be to manually tell Retrospect to eject the tape from the drive and quit, with the preference set for rewind, causing the tape to rewind. On next insertion the tape will be at BOT, and the tape won't appear as erased.

 

I think that there is another way to get the Exabyte autoloader/drive into this mode where it believes that the tape is erased, and that's to run diagnostics on the autoloader then to run Retrospect. See my 7/28/2006 suggestion for a future version enhancement here:

Product Suggestions - OS X

 

Let us know what you found, see if the above information helps.

 

Regards,

 

Russ

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...