Synthetic jobs: Media server is currently not connected to master server

[Chris_W]

We have 2 media servers that are down due to hardware failure.

As per the Netbackup scheduler our synthetic full jobs kicked in today. These jobs are determined to use either of these 2 down media servers to synthesise the images.

We have 4 servers in total that have tape drives to this VTL. So I am expecting the other 2 media servers that are up, to be assigned these jobs, but that arent.

Any ideas how I can make Netbackup stop assigning jobs to down media servers for synthetic jobs?

[mph999]

(1) Easy

In the policy, change the STU to point to a Storage unit that is 'owned' by a media server taht is available.

(20 Harder but neater 

Set up Storage unit groups, so that an alternate media server is used.

Martin

[Chris_W]

Hi Martin, currently the policy is writing to a storage group, not a storage unit. We have removed the down servers from the STG.

My feeling is that Netbackup is trying to use these down media servers to read from, not write to.

[mph999]

Ahh, ok - use:

FORCE_RESTORE_MEDIA_SERVER = <oldserver> <replacement server> in bp.conf

Or set the same in Host Properties > Master server > Restore Failover in GUI

Martin

[Chris_W]

I made the change as recommended and restarted the jobs. The issue is still there.

[watsons]

That looks like we need an "alternate read server" setting, but thinking of this... this is not in the policy schedule tab. 

If the "restore failover" does not work, what about the "media host override" in the "General Server" of master server host properties?

[mph999]

Media Host Override is the same as FORCE_RESTORE_MEDIA server, just set it in the gui, as opposed to bp.conf/ registry.

[mph999]

OK, this might work, only thing left I can think of.

For the images in qeustion on the fragment line change the media server to one that is availabel (and can access the media),

 bpmedia -movedb -m <media_id> -newserver <newservername> -oldserver <oldserver name>

Martin

[Chris_W]

We were able to start these 2 problematic servers up again. Now that the devices are back again the waiting jobs could run to completion.

This unfortunately does not help me determine what went wrong; but I think this link is near to the truth, see the row at the bottom of the table "Job delays after removing a media server":

http://www.symantec.com/docs/HOWTO56204

Thanks for all the prompt replies.

[mph999 · Accepted solution]

For some reason I thought you had deactivated the down media servers, now, I see that you didn't.  This would certainly cause delays.

This is the command ;

vmoprcmd -deactivate_host -h [hostname]

This will stop NBU trying to contact the media server, but would not automatically cause NBU to use a different media server in its place.

Martin

[Chris_W]

We didn't try that because the media servers were already marked as deactivated from the Media servers section of the Java GUI. The "deactivate" option was not available from the menu.

Doesn't the command you supplied do the same thing?

[mph999]

ahh, ok, yes it does m