Stueyb

Hi peeps,

Got a problem with veritas Backup Exec (9.1 Windows, all latest SPs running Win2K SP4)

Just recently the DLT8000 packed up and we replaced it. When the new one was put in place we started getting jobs stopping at about 40% and the error was along the lines of "Could not connect to drive. Job cancelled" Now looking in the logs it doesnt actually say which server it is failing on (theres about 8 servers with 3/4 TB data to complicate it)

Is there any way I can

a) Find out which server it is having problems with
b) Get it to ignore the error and continue.

To confuse it even more, on the job lists and selection lists, all are expandable and viewable, ie no cannot connect errors.

Any ideas?

mark_h

Can you turn the logging options up to get a full detailed log? Or look in the Restore list to see what's on that tape and see if you can tell which service is missing.

We never back up more than one server in any job on Backup Exec. Instead we have one job per server, which makes it easier to rerun any single job. We then schedule all the jobs to start at 10 minute intervals so they all run one after the other.

darlodge

I had this recently on our backup routine.

Ours only backs-up about 10GB on this Ultrium tape (rest goes else where) so it is quite wasy to work out where it failed. I just looked at the failure log and it states which server it failed against.

Once you know the Server, try removing it from the backup job detail and then see if the job completes.

Once I found the Server, I tried loads of this but all failed. I eventually removed the local remote agent and re-installed it. Be warned it needs about 2 reboots, 1 after removal and one after install for the main server to see the client.

Does it state that it can't connect to a network device or the local (SCSI?) Backup device? The error does not say for sure. Check your event logs.

One thing I MUST stress if you have not already done so, is test a restore using the new drive.

IMO Controling the backups are a bloody important job that I feel is often over looked If the worst happens, you need to feel sure your routine will work. I try to test ours every month.

Darren

darlodge

Stuey, did you solve the error in the end?

Darren

Stueyb

Yep found out what it was. It was access permissions to a machine that runs some really wierd middleware (come back corba, all is forgiven ) Basically some numpty thought it would be good to harden it, and in so hardening it knocked access to this box off for everyone except the thin client interface So when Veritas tried to connect to it via an ip it refused the connection and caused veritas to abort.

One other good thing I have learnt/implemented is several sessions on one tape (ie 1 server per session) rather than one huge job that could screw up anywhere.