SCSI TAPE: dev = 0xcd000000 Failed open - st_san_safe enabled

[Popy]

Hi All,

I have a rp-2400 box. Its is in SAN with multiple XP boxes and Tape libraries.
I am getting an error
"SCSI TAPE: dev = 0xcd000000 Failed open - st_san_safe enabled " in syslog and it is repeating. Anyway my normal operations are so far not affected.

Can anyone advice me what can be the issue ?

Regards
Raneesh Vijayan

[Steven E. Protter]

Does the machine have an scsi attached tape drive? If so, is it working.

You can check your dmesg output

dmesg

You can run cstm,mstm or X Windows xstm and test the thing with an exercize command. You'll need a blank tape that you can write to for a complete test.

Lets say you don't have a drive.

You system has at one point detected a tape drive somewhere on the san and misses it.

the drivers should be in /dev/rmt

Class I H/W Path Driver S/W State H/W Type Description
======================================================================
tape 0 0/3/0/0.1.0 stape CLAIMED DEVICE HP Ultrium 1-SCSI
/dev/rmt/0m /dev/rmt/c4t1d0BEST
/dev/rmt/0mb /dev/rmt/c4t1d0BESTb
/dev/rmt/0mn /dev/rmt/c4t1d0BESTn
/dev/rmt/0mnb /dev/rmt/c4t1d0BESTnb

rmsf command will make those files go away.

If they keep coming back then there is an issue in the SAN configuration zoning that needs to be dealt with to make the error message stop.

SEP

[Sanjay_6]

Hi,

/Quote/

This behavior occurred because the st_san_safe kernel parameter was set. When the st_san_safe kernel parameter is set, the rewind-on-close tape devices can not be used. Only a no-rewind tape device can be used, i.e. operations on a no-rewind device such as /dev/rmt/0mn will work as expected.

Any opens on a rewind-on-close device will return EINVAL and the following message will be sent to the syslog:

SCSI TAPE: dev = 0x%x Failed open - st_san_safe enabled

/EndQuote/

I would think that you should disable this parameter (set it to 0) in the kernel. Otherwise you won't be able to use the rewind device 0m/1m.. for backups.

Hope this helps.

regds

[Deoncia Grayson_1]

This was checked in regards to patches on your server:

http://www.informatik.uni-frankfurt.de/RBI/hp-service/patches/s700_800/11.X/PHKL_29899.txt

Another Quote:

When a host application is writing data to a tape, another tape command like mt(1) could be issued from a different host to the same device using a device file without the
no-rewind-on-close option (i.e rewind-on-close). This second command would issue a rewind-on-close to the drive between writes (of the first host) causing the tape toreposition to beginning of tape (BOT). The first host that is writing to the drive would have no knowledge thatthe tape was repositioned and it would then continue
writing from the BOT, effectively overwriting the previously written data.

Resolution:

This patch introduces a kernel tuneable called "st_san_safe" which has a default state of DISABLED when this patch is installed. When st_san_safe is enabled by
setting "st_san_safe=1", this forces the stape driver to fail all opens to device files which do not have the no-rewind-on-close option with the errno "EINVAL". The stape driver will also log an entry into the syslog like the following:

vmunix: SCSI TAPE: dev = 0xcd002000 Failed open - st_san_safe enabled

When st_san_safe is enabled, mksf(1M) will only create new device files with a no-rewind-on-close option. The mksf(1M) command will fail with an error indicating an
invalid option was provided in the command line arguments. With st_san_safe enabled, lssf(1M) will not fully decode the density attribute of a tape special device file.

[Popy]

Thanks guys,

There was an Ultrium drive which gone offline in a tape Library.
Your inputs really helped me to trace the issue.


Regards
Raneesh Vijayan