Re: [squid-users] More on "aio_queue_request: WARNING - Queue congestion"

On Tuesday 24 July 2001 02:45 pm, Brian wrote:
> On Tuesday 24 July 2001 10:24 am, Steve Snyder wrote:
> > To recap my config: Squid v2.4S1 (with all posted patches applied) is
> > using the aufs access type and is using the default 16 threads. The
> > cache itself is on a solid-state disk, accessed with an UW SCSI
> > interface; the filesystem is ReiserFS. The overall system is RedHat
> > v7.1 (with all RH updates applied + v2.4.7 kernel) running on dual
> > Pentium3/550MHz CPUs and 512MB of RAM. This box (and thus Squid) not
> > is not memory contrained in any way. I am using 16 redirector_children
> > (Adzapper). There aren't any CPU-intensive jobs running on this
> > machine.
>
> A solid-state disk? How much space is going to cache_dir and is this run
> as a proxy or httpd-accelerator?

It is a 512MB device that is devoted soley to Squid. The cache is handled
like this:

    cache_dir aufs /var/spool/squid 320 1 256

> In either case, does increasing the thread count to 24 or 32 make a
> difference?

I don't know; haven't tried it. It's that what the message is telling me,
that I have too few threads?

Thanks for the response.
Received on Tue Jul 24 2001 - 16:01:31 MDT