Video Screencast Help

Media servers and policies containing multiple targets

Created: 19 Jun 2013 • Updated: 17 Jul 2013 | 2 comments
This issue has been solved. See solution.

So here's the thing: 

We have media servers with DD Boost installed for Data Domain, we have MUST_USE_LOCAL set on the server the have run the nbemmcmd that tells media servers to back themselves up. This is to ensure that DD Boost de-dupe at source takes place. 

But what happens in the case of a job policy that includes two or more different servers being backed up? It is not clear to me whether a backup job/policy is  handled by one media server in its entirety, or if a different media server will take over in the case of a media server that's configured as above to back itself up.  

We are migrating from Data Protector, and some of the jobs configured on that are based on backups of specific environments and may contain 3, 4 or 5 different servers being backed up by on job or policy. It is easy enough when one policy backs up one server, but how do media servers react when the policy contains multiple target servers? 


Operating Systems:

Comments 2 CommentsJump to latest comment

Nicolai's picture

the MUST_USE_LOCAL_DRIVE directive is only valid for how the backup is executed for media servers itself.

If the storage unit you point to from the policy is not local to the media servers, the MUST_USE_LOCAL is overridden.


 If all drives are down when a backup job starts running, another drive will be assigned from the next available storage unit group. This also applies for Storage Life Cycle Policies or Storage Unit Groups.

- If the first storage unit assigned for the backups does not have any media available, the MUST_USE_LOCAL_DRIVE setting will attempt to run the backup job in the next available storage unit instead of failing the backup job with a status code 96.

And from

If the client is also a media server and theMUST_USE_LOCAL_DRIVE entry is present in the master's  /usr/openv/netbackup/bp.conf file, backups for this client must occur on a local drive. However, if all the local drives are down, the scheduler (bpsched process) will allow the backups to run on a remote media server.

I still believe the statement about MUST_USE_LOCAL_DRIVE is a global attribute is false.

Assumption is the mother of all mess ups.

If this post answered your'e qustion -  Please mark as a soloution.

SteveLaw's picture

Hi Nikolai,

Thanks for your input. In our network we are using Data Domain, so no drives are local. I am using MUST_USE_LOCAL_DRIVE on the media servers to force the media servers to back *themselves* up. 

On a previous thread - - you advised that when using Data Domain MUST_USE_LOCAL_DRIVE would force a media server to back itself up. If I have a policy to backup mediaserver1 called BK_MEDIASERVER1, I don't want mediaserver2 to run that policy - I want mediaserver1 to run it. If mediaserver2 runs the policy then there will be no de-dupe at source. ALL the data from mediaserver1 will go across the network to mediaserver2, which may de-dupe the data then but it is too late - all the data has already gone across the network. 

So my understanding is that running MUST_USE_LOCAL_DRIVE on a media server will force mediaserver1 to back itself up. Whenever backup policy BK_MEDIASERVER1 runs, I want mediaserver1 to be the media server for that policy, not mediaserver2 or mediaserver99. 

But my real question is, what happens when a backup policy contains more than one backup target server. Imagine a backup policy called DATABASE_A that backs up three database servers. And all three are also media servers. What happens then? Does one media server run the whole policy? Do they take it in turn to back themselves up in three different stages? 

I hope I've explained myself clearly, thanks for your help.