Legato/Networker Labeling Problems

[Laterite]

Greetings,

For some reason, we are having trouble labeling our tapes after replacing the incremental and full tapes.

This is what we see:
# nsrmm -m -l -f /dev/rmt/0hbn -R -b Full
nsrmm: RAP error: Tape label read: I/O error
Please load a tape to drive '/dev/rmt/0hbn'

# nsrmm -m -l -f /dev/rmt/0hbn -R -b NonFull
nsrmm: RAP error: Tape label read: I/O error
Please load a tape to drive '/dev/rmt/0hbn'

Additionally, mt status shows the Full tape to be at EOT:

# mt -f /dev/rmt/0 stat
Quantum DLT8000 tape drive:
sense key(0x13)= EOT residual= 0 retries= 0
file no= 0 block no= 0
# mt -f /dev/rmt/1 stat
Quantum DLT8000 tape drive:
sense key(0x0)= No Additional Sense residual= 0 retries= 0
file no= 0 block no= 0

When I rewind the tape, it only runs for a second, and the mt status shows the tape as ready to go at the beginning of the tape.


Auto-media Management is disabled. While troubleshooting with our vendor, we turned this feature on. nsrmmd would simply spin the tape up without ever labeling the tape, and eventually fail with an I/O error as well. We shutdown nsr, disabled auto-media management, and retried labeling with nsrmm as before. We would then see the following:

Jan 7 00:32:39 admin root: [ID 702911 daemon.notice] NetWorker media: (warning) /dev/rmt/0hbn reading: Not enough space
Jan 7 00:32:45 admin root: [ID 702911 daemon.notice] NetWorker media: (notice) Volume "Unknown" on device "/dev/rmt/0hbn": Block size is 10240 bytes not 98304 bytes. Verify the device configuration. Tape positioning by record is disabled.
Jan 7 00:32:52 admin root: [ID 702911 daemon.notice] NetWorker media: (warning) /dev/rmt/1hbn reading: Not enough space
Jan 7 00:32:56 admin root: [ID 702911 daemon.notice] NetWorker media: (notice) Volume "Unknown" on device "/dev/rmt/1hbn": Block size is 10240 bytes not 98304 bytes. Verify the device configuration. Tape positioning by record is disabled.

Had the datacenter SA remove the two tapes and replace them, and even rebooted the box to have Solaris reread device configuration in case the drive was out of sync somehow. Now, we are back the original I/O error as before.

System is a Netra T1 200, running Solaris 5.8 Generic_108528-08. Tape drive is a DLT8000, tapes are DLT Type IV. NetWorker Version is 6.0.2.Build.251.

Is this an issue with the netowrker application, or possibly the OS; i.e. does it need to be patched up to kick the tape drive into gear? I'm thinking it's the former.

Thanks,
-Mark

 

[605]

When you recycle/relabel a media, NW always tries to read the existing label first.
This fails on your system and i actually do not know why. A NW label is always written with 32kB but the block size for the data following
that label may be larger (98304 bytes for a DLT8000, but this depends on the device type).

As NW never uses 10kB blocks, it looks like the media has been (accidentally)
overwritten.

To solve the problem, simply try to label, not to re-label
the media. If
this does not work, use tapeexercise (NW) or tar to write another pattern.
Then try to label it again. This shall solve the NW problem. However,
you should investigate why the tape obviously has been overwritten and
prevent such case for the future.

 

[DavinaTreiber]

I can't see an obvious reason for your problem, however there's one thing I would check. Is the st patch up to date? To check this, do
showrev -p | grep 108725
and see what version it is. I'm not sure what rev you would need for DLT8000, but it's a good idea to have a fairly recent version. I'm also not sure what the latest is, may be 14, may be higher by now.

Also, possibly unrelated, why are you using the hbn device? This is not going to give you maximum performance or capacity. I recommend you use the cbn device to enable compression.

Reading your post again, I don't think you should have the -R flag on the nsrmm command. This is to relabel tapes that have already been used for NetWorker, I'm assuming these are foreign tapes. Just a thought...