RE: [squid-users] Questions about Squid 2.5.STABLE7 performance a nd cachemgr

From: Chris Robertson <crobertson@dont-contact.us>
Date: Tue, 16 Nov 2004 16:09:34 -0900

Wow. This sounds familiar. I had a problem that sounds just like yours.
Are you by chance using url_regex acls?

As for the active client requests, check the start line to see how long the
request has been active. If it's a big download, or streaming content, the
connection will be open for a while. It might not be such a good indicator
of problems.

Chris

-----Original Message-----
From: Tom.Williams@diversifiedsoftware.com
[mailto:Tom.Williams@diversifiedsoftware.com]
Sent: Tuesday, November 16, 2004 3:19 PM
To: squid-users@squid-cache.org
Subject: [squid-users] Questions about Squid 2.5.STABLE7 performance and
cachemgr

I've got Squid 2.5.STABLE7 running on a Fedora Core 2 based Linux box and
it's running well, except for one thing. Intermittently, there will be
delays in accessing sites and/or Squid itself. If I'm able to access the
cachemgr, the "Client-side active requests" page will show many (5 - 10+)
client requests that are waiting for a response (which I don't think is a
Squid issue). Sometimes I'm NOT able to access the cachemgr and my browser
will "spin" for a minute or two and then things will load really quickly.
When things are operating normally, Squid performance screams. :)

Any ideas as to the cause of these delays?

With regard to the cachemgr, in the DISKD section does the "unlink" count
reflect the number of cache objects that have been purged from the cache?
I would like to know how many objects are being deleted over time. :)

I've got DISKD and lru as the policy configured and the proxy is serving
about 10 - 15 machines.

Thanks!

Peace...

Tom
Received on Tue Nov 16 2004 - 18:09:37 MST

This archive was generated by hypermail pre-2.1.9 : Wed Dec 01 2004 - 12:00:01 MST