How does the SCSI driver and Linux handle tape errors ?
|
Author |
Message |
|
|
Posted : Fri, 30 May 2008 15:09:46
Subject :
How does the SCSI driver and Linux handle tape errors ?
I'm new to Linux but have a strong background in VMS administration (hey, it ain't dead yet !)
My question is, both DLT amd LTO tape drives employee read-after-write head designs with CRC firmware. These technologies can generate 2 different types of errors, soft (the CRC was able to correct an error) and hard (CRC can not able to correct the error).
Somehow, notification of these error must be sent back to the driver.
Does the SCSI driver log these errors and/or pass them on to any application (like gtar) using this device ?
|
|
|
|
Johannes Truschnigg
|
Posted : Sun, 15 June 2008 19:51:14
Subject :
How does the SCSI driver and Linux handle tape errors ?
Hello there,
I don't have particular experience with SCSI tape technology (I do backups on DVD-RAM and ordinary harddisks these days, and my (ATA) streamer was outphased some years ago already), but if the SCSI tape driver behaves similar to just about any other storage driver these days, you'll find status info about the driver's operation in the Kernel's debug ringbuffer. Any ordinary user may take a peek into it via the program `dmesg`; most distros, in addition, set their syslog daemon up in a way so that the information there which would normally be discarded once this buffer overflows is saved to a file under /var/log/.
If a serious error on the driver/Kernel-level occurs, you will notice that in userspace, too. The "how", however, I can only take guesses at:
If you're unlucky, your backup program (`tar`, for instance) might just not return at all, or (if you're slightly luckier) will return with an exit code greater than 0.
Note that there's also a SCSI debug driver ("scsi_debug.ko") which might enable you to set up a testing scenario (without actually hosing a tape or tape-drive in the process). `modinfo scsi_debug` should give you hints on how to make use of the module. Also peek around in /proc/scsi/ for status info on SCSI devices.
If you've got your Kernel's source installed, you may also dig around in the SCSI subsystem's documentation under /usr/src/linux/Documentation/scsi - I'm sure you'll find a detailed answer to your question in there somewhere.
Hope that was info you could use - keep having fun with GNU/Linux! :-)
|
|
buggy
|
Posted : Tue, 17 June 2008 11:15:10
Subject :
How does the SCSI driver and Linux handle tape errors ?
OT but if ur using tape for backup I find alternate methods work better - something like http://www.boxbackup.org/ connecting to a hosted server in a datacentre.
|