Only most recent full backup available in restore?

[EricHutchins]

Hello all,

I am running BE2010.  I recently had a user delete a bunch of data (critical, of course), and discovered that in my restore section, only the most recent full backup and incrementals prior to it were available for restore - but not the full backup from a week before.  I'm running full backups once a week and incrementals every day. 

I have a suspicion that the reason for this is that I had the retention time on the media set used for full backups set to one week.  Is this the case?  If not, what else should I be looking at to ensure that I have access to restore from my full backups not only from the most recent run but also previous runs?

I have already modified the retention time on the media set to 4 weeks, and created a device pool of several B2D volumes in the hopes of harnessing the free space that's on the three separate devices that hold the B2D folders I'm using.  

I hope this fixes it, but the bottom line is I have to be sure - the boss is going to expect to see two full backups in the restore tree come Monday.

Does anyone have some input?

[CraigV · Accepted solution]

Hi,

If your retention period is set for a week, that's how long your data will be protected for. You'd need to extend this to 2, 3, 4 weeks (whatever you require) in order to keep data for that period of time.

This would obviously mean using more tapes which you'd need to cater for if you don't have spare tapes now.

As for the data that's lost, if you don't have an older backup (like a monthly), then you've got nothing to go back too. You can search through the media set/s to see if any older tapes are available.

I really would recommend you check out policies too...specifically a GFS policy (Grandfather/father/son...or...Monthly/Weekly/Daily) if you're not using this currently. It's going to allow you to set up individual retention times for these backups. Say: daily backups kept for a week; weekly backups kept for 3 weeks; monthly backups kept for a year. It would mean more tapes, but when you create a policy, it sets up rules to disallow jobs clashing...for instance, a Monthly job would always run over a Weekly job.

Check the admin guide for more information.

Thanks!

[pkh · Accepted solution]

Your suspicion is probably correct.  One week a too short and you would end up with one backup.  There is no need to keep incremental backups for 4 weeks.  You might want to stick to 1 week for your incremental backups and create another media set with an OPP of 4 weeks for your weekly backups.  This will optimise your disk space requirement.

Since your disks are online simultaneously, you might want to want to set them to span from one disk to another.  See my article below

https://www-secure.symantec.com/connect/articles/how-span-disk

You should also make sure that your jobs specify overwrite.  Don't append to disk media because there is no advantage in doing do.

[EricHutchins]

Thanks, Craig.  I was able to find the old data, so that's not a problem (thankfully).

Now that I have adjusted the retention times on my B2D media set to 4 weeks, should I be able to see more full backups available for restore going forward?

I have trouble writing full backups to tape, as my backup set is huge, so I only write the incrementals to tape.  That means I have to retain the full backups on a large NAS, so I have to be able to restore them as well.

[CraigV]

Hi Eric,

That's correct. You'd obviously have your tapes available if you need to restore from an INCR.

Thanks!

[EricHutchins]

PKH, thanks for your advice.  The link in your post was truncated, so I was unable to read the article.

You said: "Since your disks are online simultaneously, you might want to want to set them to span from one disk to another.  See my article below" 

I have three separate volumes, each of which contains a B2D folder.  I have grouped these three volumes into a device pool, and assigned the job to a media set with an OPP of 4 weeks.  Will this work, or is the spanning that you mentioned absolutely necessary in that type of scenario?

I really appreciate your help, thanks!

[Ken_Putnam]

Hmmm, that link worked for me, but here it is again

https://www-secure.symantec.com/connect/articles/how-span-disk

One other thing to consider in your planning:

If all of your FULLs are only on disk within the same building as your server, and anything happens to the building, all your data is gone.

If your backups are too large for one tape, you need to consider purchasing an autoloader, or else get 3 or 4 or ?? external drives and write your weekly fulls to them for off site storage

[pkh]

If you click on the link it is working, Connect truncates the link because it is too long.

You can either implement spanning or choose not too.  With spanning, when you hit the low disk space threshold on one disk, the job will continue with the next disk in the pool.  Without spanning, when the job runs out of space, it will fail.