B. Cook wrote:
Running a squid box for a large network, about 3k computers; but only
about 1/3 are ever on at one time..
Squid Object Cache: Version 2.7.STABLE7
Start Time: Tue, 06 Apr 2010 17:49:52 GMT
Current Time: Tue, 04 May 2010 13:16:48 GMT
Connection information for squid:
Number of clients accessing cache: 888
Number of HTTP requests received: 23157816
Number of ICP messages received: 167211
Number of ICP messages sent: 167401
Number of queued ICP replies: 0
Request failure ratio: 0.00
Average HTTP requests per minute since start: 578.3
Average ICP messages per minute since start: 8.4
Select loop called: 614713658 times, 3.909 ms avg
Squid is a FreeBSD 8.0-p2 i386 machine with 2G of ram; bge nic with
rxcsum and txcsum enabled.
I have an occasional issue that sometimes clients will receive a 'proxy
refusing connections' message (firefox is more straight forward with the
message than IE - but the network is mostly IE)
Instantly hitting F5 or refresh causes everything to return to normal
and the user goes on, it's a 'rare' thing that happens, but occasionally
it happens.
I don't see anything obvious in cache.log..
Where/How could I start to figure this out?
With your debug_options is set to at least ALL,1. Check cache.log for
WARNING, ERROR, FATAL or "Starting". See if their times match up to the
blackout periods.
Check for any system triggered restart / reconfigure events that may be
causing Squid to reset connections. It's normal for the current releases
to have a small blackout during those.
If there is any cyclic pattern to the outages it may be one of the
internal cleanup cycles (internal roughly hourly by default for the
worst one: cache purging).
Our squid is an authenticated proxy.
We are also doing squidguard as redirect_program.
And we are using wpad autodiscovery to configure the proxy to everyone.
Thanks in advance.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE9 or 3.1.3
