Re: [SQU] Squid 2.3s4 WCCP, Reading next request

On Wed, 8 Nov 2000, Greg Holloway wrote:

> Hi,
>
> We're running FreeBSD 4.1.1 with the GRE patch, squid 2.3-STABLE4.
>
> Since installing WCCP (instead of routing all traffic via the squid PC)
> our squid CPU utilisation has doubled. About half our file descriptors
> are in "Reading next request" status, all of which have positive numbers
> in the Nread and Nwrite column. Apart from that, squid is operating
> very well with WCCP.
>
> I am wondering whether this is a squid issue or TCP-related FreeBSD issue?

sounds like you just have a lot more persistent connections for
some reason. Lowering 'pconn_timeout' may help.

> Secondly, one of our 4GB Store Directories has been allocated only
> 262000 filemap bits, whereas our other partitions (same megabyte size &
> directory structure) are allocated 1048576 filemap bits. As a result,
> squid has used all the filemap bits but only 50% disk of the affected
> Store directory. Can I adjust this without reformatting the directory
> (partition)?

filemap bits are allocated dynamically as needed. There is nothing
to worry about here.

If the disk is not getting more than 50% full, then its because
of something else. What does 'Store Directory Stats' from the
cache manager say? e.g.:

        client mgr:storedir

Duane W.

--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html