Veritas-bu

Re: [Veritas-bu] SLP with Synthetic backups - What are people doing for workarounds?

2012-02-22 09:11:22
Subject: Re: [Veritas-bu] SLP with Synthetic backups - What are people doing for workarounds?
From: Gregory Demilde <gdemilde AT gmail DOT com>
To: David Stanaway <david AT stanaway DOT net>
Date: Wed, 22 Feb 2012 15:07:35 +0100
Hi,

Basically, What I have been told by Symantec is that the full synthetic process was not aware of the SLP and therefore the image was re-created it was without the SLP information. The good workaround was to use vault duplication. Indeed that should be now fixed in one of the 7.1 releases. A year ago I think it was planned to be fixed in the 7.1.0.3 release. A simple workaround is to basically use vault duplication.

Because in 6.5.x there were some flaws in the SLP process especially when you stop and started the services, an image could get the status SLP_COMPLETED without any copies being made, I prefered to script the duplication process instead using vault so I could include the images that have been left out by the SLP. I think in 7.X Symantec corrected some issues.

The big drawback of SLP, is the queued jobs. If you do not set up duplication job size correctly, you might set up an important load on the master. Besides, depending on the size of you environment you might need to tune up NBRB. There are some SLP tuning recommendations on Symantec website. But yeah SLP queued jobs are a pain.

I do not know if it is still the case but on 6.5.x stopping netbackup master services could trigger SLP proces to be in disaster recovery mode when starting up the services. Basically, this mode forces Netbackup to check every image on the source STU before doing any SLP duplication. This can take a while ( for us 36 hours ) if you use big disks for backups.

Greg 

On Tue, Feb 21, 2012 at 4:23 AM, David Stanaway <david AT stanaway DOT net> wrote:

I am happy to report this was seemingly silently fixed in 7.1 (At least
in 7.1.0.3), the only oddity is that each backup job seems to spawn it's
own SLP duplication job, so you will see hundreds of SLP duplication
jobs queued.

Now can anyone say if SLP and GRT backups are going to work correctly in
7.1 ?

On 3/16/2010 8:02 AM, David Stanaway wrote:
> I'm getting very frustrated about having to re-invent the wheel to
> reproduce the functionality of Storage Lifecycle Policies for my
> Synthetic Full backups.
>
> For those not aware, Synthetic backup images for some dumb reason that
> hasn't been explained to me cannot be managed by an SLP and you will get
> dataloss if you backup to an SLP and expect it to execute the policy.
> The backup will go to the Backup destination, but that is it. No
> duplications.
>
> I am working on some scripts to re-produce the functionality of SLP
> since Vault is not adequate as I have hierarchical duplications that are
> not possible with a vault job. It is going to take a little time to get
> right, and I can't help but think this must have been done before.
>
>

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



--
Gregory DEMILDE

_______________________________________________
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>