Koozali.org: home of the SME Server

Affa question - full disk

Offline kevinb

  • *
  • 237
  • +0/-0
Affa question - full disk
« on: March 19, 2008, 06:26:16 PM »
This is a great contrib!

What happens when the target partition for the backup becomes full? Does the backup stop occuring? Do the oldest backups get deleted?

Also, is the disk full percentage that triggers the warning email configurable?


Thanks again,


Kevin

Offline raem

  • *
  • 3,972
  • +4/-0
Re: Affa question - full disk
« Reply #1 on: March 20, 2008, 08:14:37 AM »
kevinb

Quote
is the disk full percentage that triggers the warning email configurable?

See
http://wiki.contribs.org/Affa#Job_configuration_properties

DiskSpaceWarn    
strict or normal or risky or none    
run a disk space check after a job has been completed. With level 'strict' a warning message will be sent, if the available space is less then the size of the just completed backup. With level 'normal'/'risky' the message is sent, if less than 50%/10% of the backup size is still available.
...

Offline kevinb

  • *
  • 237
  • +0/-0
Re: Affa question - full disk
« Reply #2 on: March 20, 2008, 05:45:14 PM »
Thanks Ray,

I do not know how I missed that. I guess I was looking for a place to enter a number.

Now what happens if the disk does actually get full? Does it stop backing up or does it over write the older files?


Kevin

Offline m

  • *****
  • 276
  • +0/-0
  • Peet
Re: Affa question - full disk
« Reply #3 on: March 20, 2008, 08:39:12 PM »
Kevin,

Now what happens if the disk does actually get full? Does it stop backing up or does it over write the older files?

No, Affa will never overwrite older archives.

On every job run Affa creates a scheduled.running archive. Unmodified files are hard-linked against the latest available archive, modfied files are created new and require disk space. If the disk get full the rsync program terminates with an error. You now have an incomplete scheduled.running and no scheduled.0, but all other older archives are still there and untouched. Under normal conditions the job would have completed successfully and the scheduled.running would renamed to scheduled.0.

Please note that this is theory and not tested. I myself usually take serveral measures to avoid a disk full condition: sendStatus=weekly (and reading the mails), DiskSpaceWarn=strict, monitoring by Nagios and redundant backups (on >= 2 Affa servers).

In general, you should avoid running out of disk space in any case, not only on backup servers.