BE Can't Delete Old .bkf Files From Media

Question

Hi,We’ve had to upgrade version 2012 to 16 because of new servers we need to back up. We deployed the software on a 60-day trial on the same hardware we’ve been using with the earlier version. We back up to removable USB 3.0 hard disk drives which we use in rotation – always taking the most recent offsite. Every night the source servers are backed up onto a pair of these disks which are then disconnected in the morning and swapped for the next pair ready for that night’s backups. There are ten pairs, four on a daily cycle (Monday-Thursday) and six on a weekly cycle (Fridays). Since the upgrade we’ve encountered a problem: Backup Exec is unable to delete or overwrite the .bkf files from expired jobs on these disks.&nbsp;After the next pair of disks are connected the backup sets on them expire correctly according the retention period set for the jobs. But the disks continue to show the space as ‘used’ in the Backup Exec Storage section. That night’s backups then fail as the disk runs out of space.&nbsp;Drilling into the media shows no backup sets.We enabled the ‘Media’ view in the Registry, sometimes there are entries (always showing as overwritable), sometimes there is nothing.If the disk is disabled in Backup Exec and examined using Explorer then there are a whole lot of files in the BEData folder with the extension .bkf. If these are deleted (just the .bkf files – nothing else) and the disk re-enabled in Backup Exec then sometimes it shows most of the space now available. If not then inventorying the disk generally gets it.This is ridiculous – BE 2012 could always overwrite it’s media and it’s the same media.We’ve tried deleting the media from Backup Exec then formatting it and re-adding it as new media but the same thing happens.All the services except the ‘Agent Browser’ and ‘Management’ run as Local System and the disks are formatted using FAT.&nbsp;Ran B2DTest.exe on one of the disks and it seemed ok.The Backup Exec HCL says “any device recognized and supported by the Windows platform should work with Backup Exec” and lists only storage devices.So the bottom line is that we’re getting good backups of our servers but we’re having to micromanage all the media.&nbsp;Anyone have any knowledge or suggestions as to how to get 16 working like 2012?

pkh · Answer

Is the entire backup chain contained in each pair of disks or is it spread out over the disks that are used each day?

siemarque · Answer

The backups on one drive of each pair are full backups of several small servers - system state, and the rest. Each is self-contained and no other backup is required to recover a server. The other drive in each pair holds a full backup of the document files. Again, each drive contains an entire full backup and no other backup is required for recovery.

The media for the Monday-Thursday jobs are in two pools, one pool for the server backups and the other for the documents. Each pair contains one drive from each pool. The backup jobs are set to use the appropriate pool - not named media. The same arrangement applies to the Friday backups which use two other pools whose drives rotate on a six-week cycle.

The jobs run through the night and the drives are disabled & disconnected in the morning, then they are taken offsite. The next pair (for that evening's backups) is connected and enabled that morning. This pair of drives contains old backup sets - either a week old or six weeks old depending on which set they belong to. Then they just sit there - the backup sets in them are showing as 'expired' but that's all. After a while the sets may vanish but the media continues to show the space as in use. If they are left that way then the next backups will fail during the night as the space runs out (actually the remaining jobs are just queued waiting for media).

There doesn't appear to be any other way to make this work than go in and remove the .bkf files as outlined in the original post each day.

colin_weaver · Answer

Do you have a separate job definition for each day OR is it one job definition that repeats daily?
Also is the weekly job configured as part of the same job definition as the daily jobs, or a separate job and if a separate job is it one job defintion that repeats weekly or is it lots of job defintions, 1 for week1, 1 for week2 etc?
If you use lots of separate job definitions then you end up with multiple last backup sets which will only be reclaimed (deleted) after you have run a job from the same job definition again.
&nbsp;
If you cannot work out what is going wrong I suggest you log a formal support case
&nbsp;
&nbsp;

siemarque · Answer

For Monday to Thursday there’s a Daily Job for each server going to the Daily Server pool. There’s also Daily Data job going to the Daily Data pool.

For Fridays there’s a Weekly Job for each server going to the Weekly Server pool. There’s also a Weekly Data job going to the Weekly Data pool.

So there’re really only two definitions of each thing, one daily and one weekly. I know we could amalgamate the server definitions into a single multi-server job, and we probably will once the dust has settled.

pkh · Answer

You might want to check the BE audit logs to see whether the DLM grooming process runs after you have connected the disks for the day

Forum Discussion

BE Can't Delete Old .bkf Files From Media

5 Replies

Related Content

Unable to delete VRTSpbx log files

Deleting .bkf Files

Deleting files

Accedently deleted main.cf file

Deleting Log files