Squid 2.3STABLE1 and disk space over limit errors

From: david leigh <vorlon@dont-contact.us>
Date: Thu, 3 Feb 2000 12:42:29 +0800

Hi All

I am a very new subscriber to this list, so even though I have read the FAQ
and searched the archives I may have missed recent traffic and apologise if
this is a well covered thread.

I am having a lot of trouble with Squid after upgrading from 2.2Stable5 to
2.3Stable1. The OS platform is a reasonably standard Redhat 6.1 (few tweaks
to fs.h posix_types.h and limits.h before compilation for greater file
descriptors but thats about all) which has been running peacefully for a
long time with no real problems. The hardware is P3 450 with 512MB and
Ultra-Wide SCSI 4Gig cache volume.

The error, is that after running for a while cache.log starts reporting that
I am running out of space on the cache volume, and that it is shrinking the
cache size

WARNING: Shrinking cache_dir #0 to 94741 KB
until...
WARNING: Shrinking cache_dir #0 to 0 KB

Once that happened it now just repeats over and over again

WARNING: Disk space over limit: 21 KB > 21 KB

It looks as if squid might not be overwriting items in the cache_dir with
new ones once the cache_dir fills up?

1) I have checked the permissions and the squid process has full permissions
to the cache_dir which is a seperate hard drive mounted as /cache.
2) I have tried decreasing the size of the cache_dir to just 3Gig (Ie should
leave an unused 1gig on the 4Gig drive) and running squid -k reconfigure,
but the same problems resulted.
3) I have tried wiping out the drive, re mkfs'ing, and then beginning squid
from the beginning with a squid -z ; RunCache & but all this accomplished
was clean operation until the cache filled up and then the same problem
again.
4) The only change I made to the squid.conf file between upgrades was to put
add the "ufs" tag to the cache_dir statement
**NOTE** - I am running on a Linux both and have configured with
the --enable-async-io option but chose to not use async-fs because the
release notes said it might be buggy. Could this be the reason (I haven't
wanted to test async-fs because even though I'm getting these errors things
appear to be working - and this is a production machine).

If anyone has any ideas or comments I would be exceptionally grateful. If
there is anything else about my config that would be helpful let me know and
I'll post it.

Cheers
David Leigh
Received on Wed Feb 02 2000 - 21:39:43 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:50:55 MST