[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

missing disk space

I have several servers running on woody with a 2.4.18 kernel using ext2 file systems (yes, I know its old.) Until recently, they've been extremely stable. Then I found that the /var partition on one of them was completely full. I managed to free up about 860 MB while I went looking for the cause.

The first thing I found was that df and du did not even come close to agreeing. Here is the most recent:
home:~# df -h /var
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda2              10G  8.8G  716M  93% /var
home:~# du -hs /var
120M    /var

I also noticed that the amount of free space according to df was steadily decreasing, but the du results stayed the same. The rate is not consistent (it averages to about 1 or 2MB per hour) but the drain is constant.

At the time I discovered this, the server was running sendmail for backup mx with milter-greylist. It also was serving as a file server (with nfs, /var was not exported), a mysql server (databases are not on the /var partition) and a dns server (bind9). I moved the backup mx to another server and shut down the mail stuff. Then I commented out every nonessential cron job I could find (like mail queue ageing) and restarted as many services as I could (like syslog and ntp). Nothing seems to have helped. My other servers are having no problem.

I know I can probably just unmount, run e2fsck and get the space back. But I would like to know what the problem is first. Has anyone else seen something like this? All other mentions I found in google concerned single incidents and usually involved a power outage. I could not find a mention of an ongoing "disk leak".

Peter A. Dumpert
Innovative Computer Services, LLC
The Diamond Standard of Internet Business
P: 732-683-0092 x 102 F: 732-577-9390

Reply to: