file corruption on tape reload

Metod Kozelj metod.kozelj at rzs-hm.si
Thu Mar 14 23:46:29 PST 2002


Hello,

J. Hart wrote:

>      I have been having difficulty making a good tape backup recently.  I use
> the following command to back up a directory :
> 
>      tar cvM -f /dev/st0 2>&1 aminet | tee log
> 
> I reload that into a seperate directory using the following command :
> 
>      tar xvM -f /dev/st0 2>&1 | tee log
> 
>      When I compare the original with the reloaded backup, I find that one of
> the files in the reloaded version is corrupted.  It turns out that a contiguous
> block of exactly 4096 bytes is corrupt.  If I reload the same backup tapes
> again, the corruption shows up at a different point in a different file, but the
> corruption still involves a block of exactly 4096 bytes.

My similar experience is actually unrelated to Linux, but symptoms were
just the same. Here it goes:

We have an Alpha PC164 system, running Digital Unix 4.0D. The system had,
apart from Mylex HW RAID, also a clumsy NCR 53x810 with external Quantum
DLT4000 off it. Writing archives (using gtar) was just fine. But, the bad
thing happened and I had to read the archives. And I was geting the same
problems you described all over in random places. If I connected the DLT
to an old HP 9000/720 running HP-UX 10.20, it read the archives without
a single glitch.

We had do push really hard on our HW vendor to replace the NCR card with a
Qlogic ISP1020, which performs fine.

I believe that Adaptec cards should be fine (I have good experiences using
them), but you just might want to look into this. It might be that your
particular card is sensitive to something and another card (even same
model) would perform well. Recently we've had weird problems with Adaptec
39160 in an Alpha PC164SX (after one year of perfect functioning it
started to dislike external box with 4 SCSI disks connected to channel 1).
The problems went away after we replaced the computer's power supply with
a new one.

Another possibility would be to lower the transfer rate for particular
SCSI device (to something as low as 3.3 MB/s). If that works, it might be
your SCSI cable or terminator to blame. I beleive that's not something you
want to make permanent though.

Peace!
  Mkx

---- perl -e 'print $i=pack(c5,(41*2),sqrt(7056),(unpack(c,H)-2),oct(115),10);'


To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message




More information about the aic7xxx mailing list