Re: [SQU] squid tuning... again...

On Tue, 15 Aug 2000, Wisnu Wijayanta wrote:

> Fellow squids,
>
> I've installed Squid-2.3S4 from the everynight autogenerated tarball, on a P
> III/866 Mhz, 896 MB, and 4*18 GB SCSI-3 disks. Base platform is Linux
> 2.2.14, gcc 2.91.66, glibc 2.1.3, and I have tuned the TCP/IP parameter
> (tcp_fin_timeout, ip_local_port_range, tcp_max_syn_backlog,
> tcp_window_scaling, tcp_timestamps, tcp_sack). Squid configured with
> dlmalloc, gnuregex, underscores, and heap-replacement (LFUDA). Our main
> internet link is 2048 kpbs, and supporting for over than 1000 users.
>
> But, I've noticed that CPU usage is considerably... if not to say very-very
> high (80-90 percent), so what's wrong? I tried to patch the source code, and
> it says reversed patch applied or something. Below, I have inserted the
> general output from the cachemgr. This sever is new, so the statistics still
> changed rapidly. Any help appreciated.

It might not be a problem, but it does seem high for your CPU.
Can you also send the output of the "Event Queue" from the cachemgr?

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