running out of disk space, Linux Mandrake 6.0

Hi all, most of my machines are running Debian potato but the one on
which I'm running Squid (2.3.STABLE3) is Linux Mandrake 6.0 with the
stock 2.2.9-19 kernel.

Previously I had been running 2.3.STABLE2 (and 2.2.STABLE4 and 5 prior
to that) on one of the aforementioned Debian machines with no problems.

When I first moved it over to the Mandrake machine (while simultaneously
upgrading the Squid version) it ran out of disk space on the cache partition.
I thought maybe this happened because I tarred up my old cache and dumped it
onto the other machine (which has a bigger cache partition btw), so I deleted
the entire cache tree and re-built it, and started fresh on that machine.

Anyway, now I'm having the same problem again and I can't see what is
causing the problem.

Here is what I see in syslog:

Jul 6 15:28:01 nomad squid[13687]: Squid Parent: child process 13694
started
Jul 6 15:28:02 nomad (squid): Write failure -- check your disk space
and cache.log
Jul 6 15:28:02 nomad squid[13687]: Squid Parent: child process 13694
exited with status 1

And this is from cache.log:

2000/07/06 15:28:08| Ready to serve requests.
2000/07/06 15:28:08| diskHandleWrite: FD 5: disk write error: (28) No
space left
 on device
FATAL: Write failure -- check your disk space and cache.log
Squid Cache (Version 2.3.STABLE3): Terminated abnormally.
CPU Usage: 0.210 seconds = 0.170 user + 0.040 sys
Maximum Resident Size: 0 KB
Page faults with physical i/o: 269
Memory usage for squid via mallinfo():
        total space in arena: 2153 KB
        Ordinary blocks: 2140 KB 1 blks
        Small blocks: 0 KB 0 blks
        Holding blocks: 176 KB 1 blks
        Free Small blocks: 0 KB
        Free Ordinary blocks: 12 KB
        Total in use: 2316 KB 108%
        Total free: 12 KB 1%

Here is the output of df:

Filesystem 1k-blocks Used Available Use% Mounted on
/dev/hda1 372101 331234 21649 94% /
/dev/sda2 298101 282704 0 100%
/usr/local/squid/cache

and df -i:

Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/hda1 96256 26982 69274 28% /
/dev/sda2 77216 15347 61869 20%
/usr/local/squid/cache

This is my cache_dir line from squid.conf:

cache_dir ufs /usr/local/squid/cache 250 16 256

Why is Squid using more than its allocated 250M of disk space?

(I saw another posting about this in the list archives but that was on
Solaris and with 2.3.STABLE2 IIRC; also, that thread stopped without any
mention of a solution.)

Thanks.
Received on Thu Jul 06 2000 - 13:59:53 MDT