Re: [squid-users] cachemgr.cgi and OS do not agree

Running ReiserFS on the cache partition with the notail option?

If so, that will explain about a 5-10% increase in disk usage, over what
Squid know about. Even if not, Squid only keeps up with the objects
actually stored on the cache disk, not the metadata, padding, and
whatever else the filesystem keeps up with. So with a 4k block size,
even in ext2, you'll have a bit of empty, but 'used' space.

Stefan Berg wrote:

> Sorry for repost - but I had no response to my previous message and I really
> could use some help... :
>
> Hi,
>
> I'm experiencing a discrepancy in disk usage between what cachemgr.cgi
> (2.3STABLE4) reports and what my OS (RedHat Linux) reports.
>
> Here is what cachemgr.cgi reports:
> Storage Swap size: 11240894 KB
>
> Looking in my filesystem:
> # df -k
> Filesystem 1k-blocks Used Available Use% Mounted on
> /dev/hda1 608724 68108 509696 12% /
> /dev/hda8 17433512 13809956 2737976 83% /cache1
> /dev/hda6 806368 701292 64112 92% /usr
> /dev/hda7 396623 25848 350294 7% /var
> #
>
> So Squid says "you are using 11,2 GB", and my OS tells me "you are using
> 13,8 GB". A 2,6 GB diff!
>
> Fact is that I am also storing my squid logs on the same partition (about
> 285 MB worth) and of course there is the swap.state file (5,6 MB worth).
>
> So the difference is a little less - not 2,6 GB but rather 2,3 GB. But a
> MAJOR difference anyway.
>
> How is this possible? Today I deleted my swap.state file to make sure it
> wasn't corrupt, and I have also run fsck on the /cache1 partition without
> any problems found.
>
> Ideas? Please help me out here...
>
> /Stefan

-- 
                                   --
                      Joe Cooper <joe@swelltech.com>
                  Affordable Web Caching Proxy Appliances
                         http://www.swelltech.com