Savegroup restarts after completion.

[mliafrate]

Hello all. I have a rather large file system, about 800GB, that I have been doing incremental backups on just fine. I decided to do a full backup over the holidays while things were slow so that I could have a copy of the save set. The full backup completed and then immediately said that there were no full backups of the save set in the media database and started a full bakcup again. I stopped the backup and let the save group run at it's normal time and it did the same thing. It ran a full backup and when it completed it said that there were no full backups of the set in the media database and immediately started it again. Now I can't run incrementals anymore because there are no full save sets in the database. The backup takes a long time, usually about 3 days, so nothing else gets backed up while that save set is running. The only change that I made was to the schedule where I did an override for a full backup on 12/25/03. Any ideas would be greatly appreciated.

 

[605]

In general, there could be 2 issues:

- The configuration is bad.
Have a look at the hidden attributes of your group - maybe your
interval is set to less than 24 hours.

Another parameter (autorestart) maybe set as well.

- The media DB is bad and cannot find the full any longer.
In this case you may try to scavenge the media db or you
must run a disaster recovery to recover a valid version.

 

[mliafrate]

It seems to have corrected itslef. I ran a backup with that save set removed from the client resource and the backup ran fine for all of the save sets. Then I added that save set back to the client resource and it did an incremental just as it was supposed to do. I suspect that you were correct when you said that it may have be an issue with the database but I am not sure what it was.

 

[DavinaTreiber]

Your scenario would really worry me if I was the backup administrator. 800GB is very big for a single volume (from a backup point of view). You have already indicated that a full backup is difficult to organise, imagine the situation if you had to run a restore of the whole volume using dozens(?) of incremental backups. It would take days, or even weeks to read all those tapes, and if just one of the incremental tapes becomes damaged the restore would be near impossible. Another thing that would worry me is the possibility that some files may not be getting backed up, perhaps if their modification times are before the last full (this can happen, perhaps if files have been extracted from a tar archive or zip file). I'm not mentioning the archive bit here, because you haven't said what platform it is.

If it is at all possible, I would suggest splitting this into smaller volumes. My own preference would be not in excess of 100GB, but this partly depends on your choice tape technology.

 

[wallace88]

Run the following to do a media database check.. these will just check the database indexes and crioss check with the client file index.

nsrck -m
nsrim -X

Then I would look at the daemon.log to see if there are any obvious errors that were reported during the time you saw that backup problem, or any other time for that matter.

Third, media scavenging is another thing you can try. This procedure will compress and re-index the media database.

Fourth, test recoveries to make sure that you can recover the data. Backups are useless unless recoveries are verified.