False disk full errors

From: Chris Tilbury <cudch@dont-contact.us>
Date: Mon, 21 Sep 1998 09:19:04 +0100


Running squid 1.2b24, under Solaris 2.6. Our cache is based on a striped,
logged metadevice, created using Solstice Disksuite. It's a 12Gb (-1%
reserved) device, space allocated broadly as follows

root@bluebell [log]# df -i /proxy/cache
Filesystem Inodes IUsed IFree %IUsed Mounted on
/dev/md/dsk/d32 1536768 730087 806681 48% /proxy/cache
root@bluebell [log]# df -k /proxy/cache
Filesystem 1024-blocks Used Available Capacity Mounted on
/dev/md/dsk/d32 12260867 9230306 2907953 76% /proxy/cache

So, plenty of space and inodes.

The problem here is that the cache itself is never filling up. Squid is
configured with

cache_dir /proxy/cache 11850 16 256

But it will never reach that. Eventually, errors get logged in cache.log
that read

1998/09/20 23:42:27| diskHandleWrite: FD 29: disk write error: (28) No space left on device
1998/09/20 23:42:27| storeSwapOutHandle: SwapOut failure (err code = -6).
1998/09/20 23:42:27| WARNING: Shrinking cache_dir #0 to 9687385 KB

This is less than the cache config specifies (12134400 KB) and less than the
total space on the device (12138259 KB).

Anyone run into this problem themselves?



Chris Tilbury, UNIX Systems Administrator, IT Services, University of Warwick
EMAIL: cudch+s@csv.warwick.ac.uk PHONE: +44 1203 523365(V)/+44 1203 523267(F)
                            URL: http://www.warwick.ac.uk/staff/Chris.Tilbury
Received on Mon Sep 21 1998 - 01:23:08 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:42:04 MST