Veritas-bu

[Veritas-bu] Problems when robot was powercycled

2002-10-25 08:38:27
Subject: [Veritas-bu] Problems when robot was powercycled
From: Bruno.Bossier AT comparex DOT be (Bruno.Bossier AT comparex DOT be)
Date: Fri, 25 Oct 2002 13:38:27 +0100
This problem is solved and I want to share with you how it was solved. It
also leaves me with some new questions.

The basic problem was that when the ATL P7000 robot was powercycled,
Netbackup lost it's connection with the robot. It seemed that it was only
with the drives, because we were still able to run an inventory of the
robot. We changed some settings on the FC230 fibrebridge : we enabled
VirtualDriveResponse and FCInitiator. Furthermore, we replaced the Veritas
driver on Windows for the tapes by the Quantum driver and that finally
solved the issue. We ran some new tests and were able now to run backups
after the robot was powercycled (or FC connection lost to robot). The only
manual intervention still to do is to UP the drives in Netbackup.

This leaves me with some new questions. We had the assumption that when you
use Netbackup, you HAVE to use the Veritas drivers for your tape drives.

Is that really true or is that just a recommendation from Veritas ?

What advantage could this Veritas driver gives us in stead of the driver
from the manufacturer ?

Can we run into problems if we use the Quantum drivers in stead of the
Veritas drivers ?

Cheers,
Bruno






                                                                                
                                                  
                    "Greg Guenther"                                             
                                                  
                    <gguenthe AT csc DOT com>                  To:     
<Bruno.Bossier AT comparex DOT be>                                       
                    Sent by:                            cc:     veritas-bu AT 
mailman.eng.auburn DOT edu                                 
                    veritas-bu-admin AT mailman DOT eng.       Subject:     Re: 
[Veritas-bu] Problems when robot was powercycled         
                    auburn.edu                                                  
                                                  
                                                                                
                                                  
                                                                                
                                                  
                    10/22/2002 20:52                                            
                                                  
                                                                                
                                                  
                                                                                
                                                  





Does setting persistent bindings fix the problem on the wintel boxes ?

My understanding is that even if the adapter assigns the same SCSI target
to the devices on the bridge, NT/W2K

can still can mangle the tape device mapping.

We have several ATL libraries with FC-230 cards and DLT7/DLT8/SDLT drives.
(Brocade 3800/2800 switches and emulex LP8000/LP9000 HBA's)

Usually we can reboot the libraries with no problems.

What happens to the drive mappings when the library gets rebooted ?

Our problem is when the servers get rebooted when the library is down or
having problems.

The lovely Wintel boxes dynamically remap the tape device mappings and it
gets really ugly.

I am looking forward to going to NB4.5 and fixing that problem.

There is a fairly new firmware for the FC-230 that utilizes a virtual drive
feature.

You might want to ask ATL if that would help your problem.


Greg Guenther

_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu





<Prev in Thread] Current Thread [Next in Thread>