Mondorestore crashes while restoring

[RNSsystems]

Hi, mondorestore crashes while restoring a directory from tape.

Here is the output from mondorestore:

$ uname -a
Linux archserv 3.2.0-0.bpo.2-amd64 #1 SMP Sun Mar 25 10:33:35 UTC 2012 x86_64 GNU/Linux
$ mondorestore -t -d /dev/st0 -T /tmp/mondo
---FATALERROR--- backup-media-type not specified!
If you require technical support, please contact the mailing list.
See http://www.mondorescue.org for details.
The list's members can help you, if you attach that file to your e-mail.
Log file: /var/log/mondorestore.log
Mondo has aborted.
Execution run ended; result=254
Type 'less /var/log/mondorestore.log' to see the output log 

The directory /tmp/mondo was created before running mondorestore. The same error happens when I call mondorestore without any parameters. This server runs Debian Squeeze. The log files are attached to this ticket.

Thanks for help! Julian

[RNSsystems]

/var/log/mondorestore.log

[Bruno Cornec]

There are issues with you rtape drive to skip the OBDR headers:

running: mt -f /dev/st0 fsf 2 > /tmp/mondo/mondo.tmp.0gApcc/mondo-run-prog-thing.tmp 2> /tmp/mondo/mondo.tmp.0gApcc/mondo-run-prog-thing.err
--------------------------------start of output-----------------------------
mt: /dev/st0: rmtioctl failed: Input/output error
--------------------------------end of output------------------------------
...ran with res=512
running: mt -f /dev/st0 setblk 32768 > /tmp/mondo/mondo.tmp.0gApcc/mondo-run-prog-thing.tmp 2> /tmp/mondo/mondo.tmp.0gApcc/mondo-run-prog-thing.err
--------------------------------start of output-----------------------------
mt: invalid argument `setblk' for `operation'
Valid arguments are:
- `eof', `weof'
- `fsf'
- `bsf'
- `fsr'
- `bsr'
- `rewind'
- `offline', `rewoffl', `eject'
- `status'
- `bsfm'
- `eom'
- `retension'
- `erase'
- `asf'
- `fsfm'
- `seek'
--------------------------------end of output------------------------------
...ran with res=512

so mondorestore has no way to position itself correctly on the tape :-( Is your mt package very old ? Could you update to a more recent version. What give mt -f /dev/st0 status ?

Can you try these commands directly to see if you're able to manipulate the tape from Linux directly.

[RNSsystems]

mt is installed in version 2.11 from the current debian repositories.

here the output from mt -f /dev/st0 status:

$ mt -f /dev/st0 status
drive type = 114
drive status = 1358954496
sense key error = 0
residue count = 0
file number = 0
block number = 0

mt -f /dev/st0 fsf 2 said

$ mt -f /dev/st0 fsf 2
mt: /dev/st0: rmtioctl failed: Input/output error

mt -f /dev/st0 setblk 32768 said:

$ mt -f /dev/st0 setblk 32768
mt: invalid argument `setblk' for `operation'
Valid arguments are:
  - `eof', `weof'
  - `fsf'
  - `bsf'
  - `fsr'
  - `bsr'
  - `rewind'
  - `offline', `rewoffl', `eject'
  - `status'
  - `bsfm'
  - `eom'
  - `retension'
  - `erase'
  - `asf'
  - `fsfm'
  - `seek'

This streamer worked in an old server together with Mondo Rescue 2.2.0-rRRR and a 2.6.32-5-amd64 kernel. But I don't know the mt version which used in the last recovery process.

[Bruno Cornec]

Replying to rnssystems:

mt -f /dev/st0 fsf 2 said

$ mt -f /dev/st0 fsf 2
mt: /dev/st0: rmtioctl failed: Input/output error

So as long as that one is not solved, mondorestore won't work :-( Could it be a bad tape ? Do you have error messages through dmesg ? or on the screen ? Which SCSI controller is it ?

That thread which covers a similar issues gives some hints as well. (​http://adsm.org/lists/html/Bacula-users/2009-09/msg00277.html)

mt -f /dev/st0 setblk 32768 said:

$ mt -f /dev/st0 setblk 32768
mt: invalid argument `setblk' for `operation'

That one is less a problem.

This streamer worked in an old server together with Mondo Rescue 2.2.0-rRRR and a 2.6.32-5-amd64 kernel. But I don't know the mt version which used in the last recovery process.

Was it on a different SCSI controller ?

[RNSsystems]

The streamer is connected via SATA. It's a QUANTUM DLT-V4 In the current server, the streamer is connected via an Intel 82801IR/IO/IH (ICH9R/DO/DH) SATA controller. Dmesg and syslog don't say anything about this issue. On the screen only the input/output error is displayed.

Another tape don't work.

During weekend it's not possible for me to get physical access to the old and to the new server. So I can't connect the streamer to another SATA controller.

[RNSsystems]

I discovered a new interesting issue. If mondorestore asks to activate OBDR support and I choose "No", the restore process gets interrupted by a segmentation fault after I choose the destination directory for the restored data. I've added the mondorestore log file to the ticket.

[RNSsystems]

restore log - OBDR disabled

[Bruno Cornec]

This seg fault has been fixed during the curse of 3.0.x dev. Please retry with 3.0.3.

Concerning the other issues with mt, I don't see what I can do on my side.