Video Screencast Help

Backup files are not deleted according to the specified retention time

Created: 02 May 2013 | 8 comments
Lars J's picture

I'm running Backup Exec 2012 SP1 on a Windows Server 2008 R2, and currently backing up data to a iSCSI connected NAS device. I have set 2 weeks retention on disk for each job, but Backup Exec seems to want to keep the data forever..! For now I have backup files for more than a month, and without manually deleting the .bkf files my disk will be completely filled.

Do I have to specify somewhere else how long BE will keep the data on disk?

/Lars

 

Operating Systems:

Comments 8 CommentsJump to latest comment

Colin Weaver's picture

Are you running full and incremental backs - as the full cannot be reclaimed until the last incremental linked to the full can also be reclaimed?

Are these sets the last copy of a given set as by default they will not be reclaimed  until another backup of the same data exists

Also does this NAS ever get reported offline from the media server (if it does and the reclaim tries to happen whilst it is offline then we do have a defect)

Lars J's picture

Hi Colin and thank you for replying,

 

I'm running a weekly full backup of most of the jobs, followed by incremental backups the rest of the days of the week. I have set them to keep the backed up data for 2 weeks (both for the full and incremental job).

How do I know if the set are the last copy of a given set? The NAS has been stable and has not gone offline since the backup solution was built up.

 

pkh's picture

Your disk storage is managed by DLM.  See this blog

https://www-secure.symantec.com/connect/zh-hans/bl...

Maybe my article below will help to clarify the deletion of backup sets under DLM.

https://www-secure.symantec.com/connect/articles/w...

 

Colin Weaver's picture

Hi Lars - please familiarize yoruself with the links posted by pkh

Based on the information you have given us then your data should be kept for 3 weeks.

The reasons it is 3 weeks is that your weekly full retention is tied to the retention of the last incremental that is linked to the full and you have the same 2 week retention for your Full and Incremental sets, as such assuming your full runs on a Friday and your Incrementals run through to Thursday the full cannot be relaimed until 2 Weeks after the end of the Thursday Incremental, whcih woudl be almost 3 weeks after the full actually ran. The ful and the incremental sets from the same week should then be relaimed within 4 hours of the 2 week aniversary of the end of the Thuirdsay Incremental.

With regards the Last set - each sucessful new full (or new sequence of Incrementals that starts with a full) becomes the last set that will not be reclaimed. As such because you full's run weekly  and you have a 2 week retention (so longer than the period between the fulls), the last Full Rule should not be affecting you. If however you gacve been setting up one off test jobs etc then these backup sets wouls be counted as the last full because no later jobs form the saem job definition have been run. For this reason you can change the behaviour of this with a registry key.

Lars J's picture

OK, thank you all for the support. Colin: in order to keep the data for only 2 weeks - how should I configure the jobs? Keep the Full jobs for 2 weeks and decrease Incremental backups to 1 week?

 

Colin Weaver's picture

Yes setting the incremental to one week would work as then the reclaim would occur 1 week after the last Incremental (so approx 2 weeks after the full)

This of course does assume that your business processes allow you to only keep the incremental data for one week.

Lars J's picture

I have now decreased the retention time for the incremental backups to 1 week, but today (14th May) I still have backup files left from 20th April. Is there some way to verify whether DLM has done its job or not? Is there a command to manually do this?

 

pkh's picture

Unfortunately, no.  There is no such tool.  It could be that the old backup sets are there because they are the last restore point for some resource or some newer backup sets are dependent on them.