Ultrium failures using SAM and Fbackup

[Jerry Jaynes]

I posted a previous note on this, but it seems to be lost...

I have a problem with an Ultrium LTO tape drive HP-UX 11.00, and SAM.

When backing up using SAM, I get random failures, "Error while writing data record xxxx". We have spent months working onthis and have gone through the gambit with support. We have changed hardware, and even reloaded the OS. I even loaded aomniback and it that works fine, so it isn't the hardware. The failures were over 60% of the backups. TOTALLY UNACCEPTABLE!!!

I finally late last week got an email from the engineer at HP that said they are getting reports of random failures using SAM and fbackup... No comment...

The errors seem to be related to scsi timing, or beffering. I get a cryptic note in the syslog about scb->cdb xx xx xx xx xx xx

I am hoping that someone else who is having this failure can assist. I have gone to running fbackup as a command line with a cron job to get backups.

has anyone ever seen this? Can you recommend kernel configurations? I am at a loss and am not getting a lot of answers.

[steven_19]

I have received these errors even during a manual fbackup using the command line.

I have been informed that it is simply down to the fact that the attempted file/s to be backup up are being written to/used at that specific time

To double check this you could have 2 seperate sessions open. One performing the backup with the -V option so you can see what is on the screen/displaying errors. The second running sar -d 2 10 to match wether the file displaying the errors belongs to the disk displayed by the sar command.

Regards

Steve

[steven_19]

Create a text file (process) containing the following

ksh
for i in $( fuser -c /var 2>/dev/null)
do
ps -fp $i | grep $i
done

This will list all active proccesses & PID's on /var. You can then investigate where they are writing.

The stop them before you do your next recover

[steven_19]

Create a text file (process) containing the following

ksh
for i in $( fuser -c /var 2>/dev/null)
do
ps -fp $i | grep $i
done

This will list all active proccesses & PID's on /var. You can then investigate where they are writing.

The stop them before you do your next backup

[Jerry Jaynes]

D I just let it run all the time, execute it before a backup or what? Can you give alittle guidance on this?

Thanks!

[Sanjay_6]

Hi Jerry,

Take a look at the thread below. Make sure you have these patches,

http://us-support2.external.hp.com/cki/bin/doc.pl/sid=f5612d1302fd1c906f/screen=ckiDisplayDocument?docId=200000056385835

And this patch too,

http://us-support2.external.hp.com/cki/bin/doc.pl/sid=f5612d1302fd1c906f/screen=ckiDisplayDocument?docId=200000055237110

hope this helps.

Regds

[Jerry Jaynes]

Yup, patched to the hilt.