Netbackup seeding and Server refresh and data migration

[yobole]

Hi Guys

Scenario

We have a remote server without the best link backing up data across the WAN from two drives (E and F) to a centrally located Media server 1 in a policy configured to backup up all local drives to a MSDP1 pool on Media server 1

Due to some data reorganization the F drive stopped completing  hence the policy was changed to backup just the E drive with Netbackup and a DR Robocopy job was put in place locally for the F drive . The E volume is backing up successfully to Media server

• We have now installed an additional  Netbackup appliance 5230 as an Additional Media server with MSDP2 pool
• We have a copy of both the remote site E and F volumes from 3 weeks ago on an eternal drive in the central location .( There is no way we can get a more up to date copy )
• We now have to replace the remote server and would like to tidy up the backup situation on this site

End Result Required

What’s the best way to proceed to get an end result of just 1 policy backing up both the E and F volume to either MSDP1 or MDSP2? (Preferably MSDP2 but easiest pool will be adequate) from the newly due to be replaced server?

[RiaanBadenhorst · Accepted solution]

Hello

Not 100% clear on the plan/requirement but let me try.

You have server A in remote backing up to MSDP A in central. E was working fine up till now, F might be a bit out dated or the data could have maybe expired (not ideal).

If the data has not expired you can just run the backup again. Having dedupe should mean you send very little data across the wire (depending on changes of course, and mostly for F) .

If the data has expired you could plug the data into a data center client, back it up, and then use the seedutil to seed the  remote client's data.

[Andrew_Madsen · Accepted solution]

For any of what I am going to write to work I have to assume that you are using client side deduplication.

There are a couple of ways we can do this. If I follow your email the "E" drive is up to date on MSDP1 and you have a 3 week old copy of the "F" drive on a USB drive located in the data center that both MSDP1 and MSDP2 are located.

The first way is to just let the backups go and hope they finish eventually. - BOGUS. Especially if you have something fairly recent on hand. Which leaves "seeding" your backups. There is a good article on that here http://www.symantec.com/business/support/index?page=content&id=TECH144437 basically this is what you need to do.

Setting up the "E" drive for remote backup you need to identify the last full or manually run a full to MSDP1. Once that completes run bpduplicate (bpduplicate -dstunit MSDP2 -backupid imagename). This will get the latest full backup image to MSDP2. Verify this by doing a listing on MSDP2:

ls -l /disk/databases/catalog/2/ and see if the client name appears. If so look deeper and note the policy name. Hold onto that information.

For the "F" drive the basic idea is to attach the USB drive to another Windows server and back it up with NetBackup to MSDP2. I suggest using a special policy for that. 

Here comes the fun part. It would be simpler to add a line to the clients pd.conf but you have at a minimum two policies you need to work against and the   FP_CACHE_CLIENT_POLICY declarative does not support a dual target. So we need to use the process that creates a seed directory. Running this command from your 5230 it is /usr/openv/pdde/pdag/bin/seedutil. You would run it twice using the same client name.

/usr/openv/pdde/pdag/bin/seedutil -seed -sclient <client name> -spolicy <policy name> -dclient <destination client name>

The first time using the "E" drive backup and schedule and the second using the "F" drive client and schedule. After completing that run your full backup form the remote site and it should deduplicate as the backup is going. After the backup runs remove the seed directory with the -clear command.