Hi
I think that you are hitting problems with "open/close" latency.
Squid-1.2 with threads might be the only option. For more info look
at www.linux.org.za/squid/ for more info on squid-1.2
Thanks to Michael O'Reilly for my original help with this:
Find your squid-PID.
run:
strace -oz -T -p <squid-PID>
wait a couple of minutes.
run:
./elapcount < z
Here is elapcount.pl:
--------------------
#!/usr/bin/perl
while (<>) {
/^([a-z0-9]*)\(.*<([^>]*)>$/ or (print($_), next);
$diff = $2; $syscall = $1;
$map{$syscall} += $diff;
$count{$syscall} ++;
$total += $diff;
}
printf "%20s %5s %12s %12s %2s\n",
'System call','Count','Total time','Time per call','% total';
foreach $i (sort { $map{$a} <=> $map{$b} } keys %map) {
printf "%20s %5d %12.4f %12.6f %2.5f\n",
$i, $count{$i}, $map{$i}, $map{$i} / $count{$i},
$map{$i} / $total * 100;
}
print "Total time elapsed: $total\n";
--------------------
You will get some kind of info on what is causing squid to slow
down like it is... it'll probably be "close(), open()" and various
others to a lesser degree..
If it's the above, try giving squid-1.2beta20 a go... it's really stable
once you get the threads library sorted.
> 30 MB of RAM in squid.conf, I do plan to add RAM to the system but didn't
Try lowering that...
Also, try remounting your FS with "no_atime":
[root@cache2 /root]# grep atime /etc/rc.d/rc.local
echo "turning noatime support on"
mount -oremount,noatime /dev/md0
Oskar
--- "Haven't slept at all. I don't see why people insist on sleeping. You feel so much better if you don't. And how can anyone want to lose a minute - a single minute of being alive?" -- Think TwiceReceived on Tue Apr 28 1998 - 08:32:22 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:39:58 MST