Help on Tape Backups / Disc Space
freebsd-questions-local at be-well.ilk.org
Tue Feb 7 07:48:24 PST 2006
"Graham Bentley" <gbentley at uk2.net> writes:
> Can any one comment on the below ;
> candle# df -H
> Filesystem Size Used Avail Capacity Mounted on
> /dev/ad0s1a 260M 37M 202M 15% /
> devfs 1.0k 1.0k 0B 100% /dev
> /dev/ar0s1d 116G 96G 11G 90% /data
> /dev/ad0s1e 260M 29k 239M 0% /tmp
> /dev/ad0s1f 37G 934M 33G 3% /usr
> /dev/ad0s1d 260M 5.4M 234M 2% /var
> /data is at 90% capacity - its a mount of two discs
> in a raid. The rest of the os install is on a sinlg disc.
> I am using flexbackup, a perl backup script in the ports
> to do backups of /data however the log shows that it
> halts almost immediatly.
> I also notice this at the end of dmesg.today
> pid 90729 (smbd), uid 65534 inumber 2119778 on /data: filesystem full
> pid 90729 (smbd), uid 65534 inumber 2921022 on /data: filesystem full
> pid 90728 (smbd), uid 65534 inumber 1931544 on /data: filesystem full
> pid 90728 (smbd), uid 65534 inumber 1931545 on /data: filesystem full
> pid 90753 (smbd), uid 65534 inumber 1931545 on /data: filesystem full
> I was wondering if flexbackup was trying to use /data also for
> temp spooling of the backup job? /usr is virtually unused so it
> seems to make more sense to use that.
Are you just guessing here, or do you have a reason to think this is
happening? A quick look at flexbackup makes me think that it uses
virtual memory, not file space, to buffer data for spooling. Do you
even know for sure that your backup was running at the time that the
filesystem full messages were generated?
> /data was up to 98% but we removed alot of stuff of it. Are we still too
> close to full capacity ?
Maybe. At some point you filled /data up. You don't have enough
information here to indicate why. You will probably want to keep
closer track to understand your usage patterns better.
More information about the freebsd-questions