cancel
Showing results for 
Search instead for 
Did you mean: 

Full + incremental sets not being groomed by DLM

M_Strong
Level 4

Good Day,

I'm encountering an issue with Full + Incremental backup sets not being groomed properly.  I understand the logic behind DLM and I've been using BE for over 10 years.  Typically focus on full backups only, however due to infrastructure size and storage constraints during a growth spurt for the company, we were forced to implement a weekly Full + Daily Incremental backup strategy for our Exchange server:  Full runs every Sunday and Incrementals run Monday - Friday.  Every individual backup job is B2D2T (Backup to Disk and Duplicated to Tape), so I have both the full and the week of following incrementals duplicated.  Using this strategy, I have enough B2D storage for about 2 weeks of history on-disk (not to mention the copies on tape that I rotate out) - 2 Full and 10 or so Incrementals.

With this strategy in mind, one would think that DLM would groom out the first Full and first week's worth of Incrementals after the second Full backup is performed successfully, verified and duplicated to tape.  This, however is not occurring.  The B2D targets are filled up with 2 weekly Full backups + 2 weeks (10+ days) of Incermentals and when week 3 begins and the 3rd Full backup begins to run, it hangs with storage errors stating that there is not enough disk space available.  I have tried manually expiring the oldest set of backups (The first Full and first weeks of Incrementals), however BE still does not groom the files.  So, I am forced to perform a hard delete on the older backup sets and snapshot folders, (which allows the week 3 backup to continue/complete), then perform an Inventory/Catalog on the B2D volume which removes the older (already deleted) backup sets from the Storage > Backup Sets view in the BE GUI.

I have the data retention for those backup sets configured to a low value (only a few hours or days) which allows the backup sets to expire long before I need them groomed and I rely on the fact that BE will not delete the last avaialble backup set (and dependants) required to perform a restore.  This works correctly.
I have the global storage settings to: Overwrite Recyclable media before scratch    and   Allow Backup Exec to delete all expired backup sets is [OFF].

So as far as I can tell, the behavior by design should be that BE should delete the First weekly Full backup and dependant Incremental backup sets at least once the second weekly Full is completed/verified (which would be a successful SDR backup).
At the very least, after the first and second Weekly Full backups fill up the volume, when the 3rd weekly Full backup is run and encounteres a low storage condition, DLM should groom out the first full backup and subsequent first set of dependants - which it does not.  Ideally, it should not even get to the point of needing the low disk space condition since DLM should delete the first Full + Incrementals after the second Full is complete.

Can someone point me in the direction of what may be missing here or if there is some other flaw/bug that is causing this behavior?

3 REPLIES 3

pkh
Moderator
Moderator
   VIP    Certified

You have not stated the retention period that you used and that has an effect on the backup set retention.

You might want to read my article below

https://vox.veritas.com/t5/Articles/When-backup-sets-are-deleted-under-DLM/ta-p/811254

Thank you for the rapid resposne.  If you'll please note, the 3rd paragraph in the original post:

"I have the data retention for those backup sets configured to a low value (only a few hours or days) which allows the backup sets to expire long before I need them groomed and I rely on the fact that BE will not delete the last avaialble backup set (and dependants) required to perform a restore.  This works correctly."

If you'd like specifics: The weekly Full (Runs Sunday evening) retention is for 20 hours, so by Monday evening those backup sets retention periods have expired; the Weekday Incremental retention is for 1 day and no Incremental runs on Saturday.  So, my Monday - Friday incrementals have expired retention periods by Saturday evening.  No backup runs on Saturday, so by Sunday evening when the next Full back up is scheduled to run, the entire previous weeks of backups (Full+Incrementals) have expired retention and should not be blocking the grooming process.

So, the backup sets created with this strategy expire no later than 24 (to 26) hours after the backup job runs.  Given that I am accumulating 2 weeks of backups before I run out of B2D target space, at the VERY LEAST, the first week should have been groomed by the start of week 3.  Either by nature of the fact that a second FULL (SDR Success) backup has been performed or when the B2D volume becomes full, the first Full backup (and subsequent incrementals) should be groomed.

pkh
Moderator
Moderator
   VIP    Certified

Is your disk storage a fixed disk or a movable one?  If it is the latter, try leaving it online for at least 1 hour.

Did you check the audit log to see whether the grooming process is being done?

You might also want to turn off the duplication for a week to see whether this will solve the problem.  I remember some other users having problems with grooming when they duplicate their backup sets.  Can't remember the exact nature of their problems.