make_tape_recovery ---- I/O error

[Rgomes]

Hi All,

Although hpux can detect tape drive, make_tape_recovery gives the I/O error ( attachment).
# ioscan -fnC tape
Class I H/W Path Driver S/W State H/W Type Description
==========================================================================
tape 0 0/1/1/0/1/0.2.0 stape CLAIMED DEVICE HP C5683A
/dev/rmt/0m /dev/rmt/0mnb /dev/rmt/c2t2d0BESTn /dev/rmt/c2t2d0DDSb
/dev/rmt/0mb /dev/rmt/c2t2d0BEST /dev/rmt/c2t2d0BESTnb /dev/rmt/c2t2d0DDSn
/dev/rmt/0mn /dev/rmt/c2t2d0BESTb /dev/rmt/c2t2d0DDS /dev/rmt/c2t2d0DDSnb
#

Is this hardware not ok?

TIA,
Richard

[Tom Ward_1]

It looks okay from the ioscan, here's some things you can try:

1) make sure tape is not write protected. It happens.

2) try a differnet tape. How old is the one you're using?

3) Use stm (mstm, xstm) to run diags on the tape drive with a scratch tape. It it passes diags, it's not the drive.

Tom

[G. Vrijhoeven]

Hi,

In addition to the comments Tom mad i would like to add, try using a cleaning tape for you drive. If cleaned you can try writing something to the drive ( after inserting writable tape with the right format ( DDS1-2-3-4)) with the tar command.

#tar cvf /dev/rmt/0m /etc/hosts
# mt -t /dev/rmt/0m rew
restart your make_recovery session.

Gideon

[john korterman]

Hi,
it says that vgdisplay -v was unsuccessful. I assume that command has exited with a return code that was not 0, which indicates one or more errors in your volume group configuration. Try running vgdisplay -v manually; check the return code and check for errors. It is a little difficult to tell what to look for, but if this command:
# vgdisplay -v | grep stale
returns anything, then post the output.

regards,
John K.

[T. M. Louah]

More details from the make*recovery error could be helpful, when the error happens? in the very begining ... is /var or /var/tmp or /tmp filled up. What version of IUX you are using, make*recovery calls pax to execute writes, one of pre-requisites of IUX tools is to put the latest pax patch.
Make sure the tape is readable & wirteable simply by tar, but first run:

# lssf /dev/rmt/0m this should display the right info stape is the driver card instance .. AT&T best density then HW path, make sure the hw path is not replace by ??? cause that means it not correct
now send one file to be written to tape
# tar cvf /dev/rmt/0m /etc/hosts
if it works then make sure you can view tape content:
# tar tvf /dev/rmt/0m
Here is another Tip, make sure all filesystems listed in /etc/fstab are mounted properly, keep only the ones that should be mounted, check vg00 for any unused lvol, remove unused ones.

good luck
T+, pliz assign points!

[Bill Hassell]

You have unrelated errors: vgdisplay shows that something is wrong with VG /dev/vgunicorn (several disks are no longer present). Since you are backing up vg00 only, that error will have a minor effect on the vg00 backup.

The second error is a tape error and most likely you'll see tape errors logged in syslog (unless you disabled the online diagnostics). Use a simple tool like tar to verify that the tape has a problem. make_tape_recovery does not have much in the way of error recovery for the tape device.

[Michael Tully]

The LIF problem points to a tape media problem. Clean the drive and try a new tape. As suggested by Bill, try writing to the tape using tar.

(I don't wish to play points policeman, but your point assignment of 17 from 188 is terrible. People are less likely to assist you when they see these type of stats, see the forum etiquette)

[Steven E. Protter]

Ignite will not complete if there are any disks listed in /etc/lvmtab that are not visible to a pvdisplay command.

I'm attaching a script that will help with that.

SEP

[Rgomes]

Hi Michael,

Recently I am giving points to people who are assisting me. Before that, actually I could not assign points as I did not see the button for assigning points. But anyway, thanks for reminding me!

Here is the outputs:
# mt -t /dev/rmt/0m rew
# lssf /dev/rmt/0m
stape card instance 2 SCSI target 2 SCSI LUN 0 at&t best density available at address 0/1/1/0/1/0.2.0 /dev/rmt/0m
# tar cvf /dev/rmt/0m /etc/hosts
a /etc/hosts 2 blocks
Tar: end of tape
Tar: to continue, enter device/file name when ready or null string to quit.
/dev/rmt/0mn
User opened file /dev/rmt/0mn to continue.
Tar: error on new volume
Tar: end of tape
Tar: to continue, enter device/file name when ready or null string to quit.

User entered a null name for next device file.
# tar tvf /dev/rmt/0m
Tar: blocksize = 0; broken pipe?
#
#vgdisplay -v vg00 | grep stale

I should not worry about 'vgunicorn' as it is cluster vg and running on another node.

Thanks for your reply,
Richard

[Rgomes]

Hi All,

Problem is solved after changing the tape drive to another BAY( bay 4, before it was at bay 2). I guess the problem is in SCSI interface or in termintor.

Thanks a lot,
Richard