On Wed, October 24, 2007 9:04 am, Roger Searle wrote:
> Hi, I have a newish suse 10.3 box acting as a file server that I ssh
> into to check for data backups each morning.  Yesterday it was extremely
> slow to accept the password (maybe 2 minutes) and I then see that the
> boot partition that normally has plenty of space is full:
>
> [EMAIL PROTECTED]:~> df
> Filesystem           1K-blocks      Used Available Use% Mounted on
> /dev/md0              19228180  19226164         0 100% /
>
> This happened a couple of weeks previously, a reboot freed up the
> space.  I am trying to discover what has happened and rectify all via
> command line without resorting to turning on the monitor still attached
> and using KDE tools to find out what is filling up the partition.  And
> also to prevent the situation in future.  My immediate thought is what
> might be happening in /tmp but I don't know enough commands to see where
> the file hogs are - do I have lots of little files filling up logs?  or
> a few very large files?  A listing of /tmp doesn't appear to give me the
> size of the contents of sub-folders.
>
> jupiter:/tmp # ls -al
> total 96
> drwxrwxrwt 17 root        root  20480 2007-10-24 08:45 .
> drwxr-xr-x 23 root        root   4096 2007-10-15 13:42 ..
> drwx------  3 beagleindex root   4096 2007-07-24 16:00
> .beagleindexwapi.jKxRZ12051
> drwx------  3 roger       users  4096 2007-10-15 16:00 gconfd-roger
> drwx------  2 root        root   4096 1970-01-01 12:00 gconfd-root
> drwxrwxrwt  2 root        root   4096 2007-10-16 09:03 .ICE-unix
> drwx------  2 roger       users  4096 2007-10-17 12:21 kde-roger
> drwx------  2 root        root   4096 2007-10-16 09:04 kde-root
> drwx------  3 roger       users  4096 2007-10-23 12:08 ksocket-roger
> drwx------  2 root        root   4096 2007-10-16 11:19 ksocket-root
> drwx------  2 roger       users  4096 2007-10-15 16:00 orbit-roger
> drwx------  2 roger       users  4096 2007-08-23 09:37 ssh-dxsfO3874
> drwx------  2 roger       users  4096 2007-10-15 15:05 vmware-roger
> drwxr-xr-x  2 root        root   4096 2007-10-15 16:00 .webmin
> -r--r--r--  1 root        users    11 2007-10-15 13:44 .X0-lock
> drwxrwxrwt  2 root        root   4096 2007-10-15 13:44 .X11-unix
> -rw-------  1 root        root     52 2007-10-15 16:00 xauth.XXXX1Ux4oT
> -rw-------  1 root        root     52 2007-10-16 09:03 xauth.XXXXBszOCc
> drwx------  2 root        root   4096 2007-09-18 11:05 YaST2-24051-LcDiM5
> drwx------  2 root        root   4096 2007-09-18 11:05 YaST2-24051-vZdQh1
>
> Maybe I'm off track with this?  This is the same box that had the time
> drift issue until I put the "ntp restart" every 10 minutes into cron.
> Is this filling up some log somewhere?  Any pointers greatly appreciated.
>

Try du --max-depth=1 |sort -n

That will show you what directories seem to have a lot in them. Drill down
into the ones that seem to have an inordinate amount. Its best done as the
root user to make sure it can read all the files and count them. And by
the way thats your root partition, not your /boot partition :-)

On startup try running ntpdate to get the time correct before starting the
ntp daemon. The daemon will only move the time by small amounts.

logs are in /var/log and can often fill up, although a reboot would not
usually fix that. Some systems clear out /tmp on reboot, so if rebooting
fixes it, this could indeed be where its filling up.

Now you can see why server admins like /var (or even /var/log) and /tmp to
be on separate partitions!


> Cheers,
> Roger
>
>


-- 
Nick Rout

Reply via email to