Tape drive problem

[Vikas_2]

Hi,
I have one D class server runnig on 10.20. It has one DLT4000 attached to it. The moment I try to run any backup on this drive, it becomes unaccessible..ioscan shows no_hw. But after a power recycle of tape drive, it used to show as claimed. I suspected the faulty tape drive and it has been replaced now. But still the problem remains the same....I'm very much hesitant to reboot the server because I don't have any backup and it has not been rebooted for long time. Tape drive was also replaced while system was up.

Is there any other diagnostic I can do without disturbing the server.

Here is the ioscan output for this card/drive
ext_bus 1 8/8 c720 CLAIMED INTERFACE GSC add-on Fast/Wide SCSI Interface
target 6 8/8.3 tgt NO_HW DEVICE
tape 2 8/8.3.0 stape NO_HW DEVICE Quantum DLT4000
/dev/rmt/2m /dev/rmt/2mn /dev/rmt/c1t3d0BEST /dev/rmt/c1t3d0BESTn
/dev/rmt/2mb /dev/rmt/2mnb /dev/rmt/c1t3d0BESTb /dev/rmt/c1t3d0BESTnb

[Helen French]

Is that drive properly terminated? Do you have any error message on syslog.log or root mail? Is the system updated with all latest patches? Check the drive status (when it's CLAIMED) with:

# mt -t /dev/rmt/2m status

Also, check the drive with STM tools.

[Steven E. Protter]

You have a chicken or the egg problem.

Its pretty hard to take a backup without that tape drive working right.

You should check that the drive and cables are connected properly. You can do a cable swap without bringing down the server. Check the termination on that tape drive. It may be one of the new self terminating drives and might not like having a terminator on it.

You need to distinguish a total backup from your data.

You should bring down your applications and do a cold backup of your data to another server, as soon as possible.

This limits the risk when you actually reach the conclusion that this server will eventually need to be booted to complete diagnosis of this problem.

If you have enough disk space, you could use mirror/ux to mirror critical data or the bood disk to improve reliability.

Another possbility is Ignite make_net_recovery. I'm not sure if thats part of Ignite for 10.20, but it would let you back your system up across the network.

SEP

[A. Clay Stephenson]

First thing to check is that the Green LED on the back of the SCSI controller is on. If so then at least terminator power is supplied -- although this tells you nothing about termination itself. Make absolutely certain that the bus is terminated in exactly two places --- on the ends of the bus. It is sursprising how well SCSI will work with improper termination -- especially on short cables. It works just well enough to drive you crazy. Unfortunately, the only way to confirm that the controller is terminated properly is to open the box up and see if the resistor packs are in the "real" or "dummy" sockets.

[GK_5]

Vikas,

Do a scan for tape drive through SAM. Sometime when you scan the hardware through SAM it finds it and updates driver.
Are you trying Omniback backup or fbackup? Try tar also. If you have to reboot the box, I would suggest doing Ignite bakcup first. If Ignite is not installed then can be downloaded from
http://www.software.hp.com/portal/swdepot/displayProductInfo.do?productNumber=IGNITEUX

Good Luck

-GK-