cancel
Showing results for 
Search instead for 
Did you mean: 

Robot issue post upgrade to netbackup 8

Slartybardfast
Level 5

Good Evening Forum experts,

I have run in to a bit of a situation post upgrade to version 8. A little background first. I performed a hardware migration from an old ageing DL585 on 2k3 running version 7.1.0.1 to DL380 G5 running 2k8R2. All was going great up until the upgrade from 7.7 to 8.0. Once version 8 was installed I started experiencing issues with the tape library, which is an equally ageing ADIC i2000. It started reporting error 98 and varius other errors which I cannot remenber now as I think I have lost the plot a little. The main issue is that it will position the first tape and start writing then any subsequent loading of other tapes it requests then but they never get mounted for writing. The last test I did was send 4 backups to Netbackup the first tape loaded and started but the subsequent 3 other tapes never got mounted. I could see them being request in the details pane of the job, but when I queried the tape library using robtest and using the "s d" command I could see the first tape in the drive but not the other 3. All other drives remained empty. The command "tpconfig -d" showed all drive with a status of UP on the 4 paths. I have deleted the STU and tape drive a few times as on closer inspection I noticed some paths were not correct. I beleive they are all ok now. What other troubleshhoting can be done, are there any logs that might reveal whats going wrong. I am currently at home so I cannot post any outputs. I was hoping that I have just missed something obvious during the upgrade. if I have to roll back wiill I have to go through the whole restoring to version 7.1.0.1 and step forward to at least 7.7. Your thoughts and guidance greatly appreciated. Thanks in advance

1 ACCEPTED SOLUTION

Accepted Solutions

mph999
Level 6
Employee Accredited

Firstly, is the library listed as compatable in the compatability guide with 8.1 ...

That aside, you have done some good troubleshooting, we know for certain that the tape is not in the drive.

So, what does robots log show :

You should very much like you know what you're doing, but just in case ....

https://vox.veritas.com/t5/Articles/Quick-Guide-to-Setting-up-logs-in-NetBackup/ta-p/811951

I'd set up ltid, robots, daemon logs, with VERBOSE in vm.conf and DRIVE_DEBUG + ROBOT_DEBUG touch files (restart media manager to get that lot to work).

We should then see something in the robots log, like this:

22:47:47.404 [10465] <5> tldcd:move_medium: TLD(0) initiating MOVE_MEDIUM from addr 256 to addr 4096

This is the point where NBU has 'sent' the scsi cdb to request the tape be moved from slot -> drive

If we don't see such a line, we didn't get as far as sending the command, if we do, we sent the command (the line is written to the log after the command is sent).

It may be that the robot receives the command and sends back some error, or it receives the command and just does 'nothing' - or it doesn't receive the command although we sent it ...

The only real way to know, is presuming for the moment we do send the command, is to get the hardware vendor to assist in seeing what the library receives (not pointing the finger here, just requesting assistance).

 

View solution in original post

3 REPLIES 3

mph999
Level 6
Employee Accredited

Firstly, is the library listed as compatable in the compatability guide with 8.1 ...

That aside, you have done some good troubleshooting, we know for certain that the tape is not in the drive.

So, what does robots log show :

You should very much like you know what you're doing, but just in case ....

https://vox.veritas.com/t5/Articles/Quick-Guide-to-Setting-up-logs-in-NetBackup/ta-p/811951

I'd set up ltid, robots, daemon logs, with VERBOSE in vm.conf and DRIVE_DEBUG + ROBOT_DEBUG touch files (restart media manager to get that lot to work).

We should then see something in the robots log, like this:

22:47:47.404 [10465] <5> tldcd:move_medium: TLD(0) initiating MOVE_MEDIUM from addr 256 to addr 4096

This is the point where NBU has 'sent' the scsi cdb to request the tape be moved from slot -> drive

If we don't see such a line, we didn't get as far as sending the command, if we do, we sent the command (the line is written to the log after the command is sent).

It may be that the robot receives the command and sends back some error, or it receives the command and just does 'nothing' - or it doesn't receive the command although we sent it ...

The only real way to know, is presuming for the moment we do send the command, is to get the hardware vendor to assist in seeing what the library receives (not pointing the finger here, just requesting assistance).

 

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Excellent advise from @mph999 as always.

Just one thing that I've noticed in your post:


wrote:

.....when I queried the tape library using robtest and using the "s d" command I could see the first tape in the drive but not the other 3. 


When you use robtest, you are taking over robot control. NetBackup can no longer move tapes around.

I totally understand that you are trying to see what is physically happening (or not happening), but using robtest is making things worse. 
If your library has a web interface, then rather use that. 

Extract from: https://www.veritas.com/support/en_US/article.000026395

Please note that this test cannot be run while VERITAS NetBackup (tm) is attempting backups, restores, duplicates, or any other actions that involve the robot and drives. 

 

Many thanks to mph999 and Marianne,

I have marked mph999 reply as the solution as you put me on the right track with the logging which revealed that the pbx was blocking commands in the robot log. All that was needed for to pbx to be repaired, this also gave a little a bit of an issue as you cannot run Netbackup installer in the repair mode as the web credential are not prompted for and causes the software to roll back. You can how ever run the installer in install mode and repair it that way. Thanks Marianne for reminding me not to use robtest when backup and duplications are in progress. Once again thanks I can now have my half term week of in peace. :)