Re: [squid-users] WCCP becomes unusable

Adrian & Gergely,

After I replaced the 2610 to be 3620, I found the CPU decrease for around
25%. This makes my WCCP run normally. I will monitor this condition for 1
week and inform both of you.

Thx & rgds,

Awie
----- Original Message -----
From: "Adrian Chadd" <adrian@squid-cache.org>
To: "Gergely Madarasz" <gorgo@sztaki.hu>
Cc: <squid-users@squid-cache.org>
Sent: Thursday, February 14, 2002 10:45 PM
Subject: Re: [squid-users] WCCP becomes unusable

> On Thu, Feb 14, 2002, Gergely Madarasz wrote:
>
> > I don't know exactly how wccp works, but from what I read from the squid
> > wccp code it seemed to me, that squid doesn't check anything in the
> > iseeyou messages except for the change member, so how could it see that
> > its out of the cache assignment?
> >
> > And how should the cache resubmit itself? I only see that it sends
hereiam
> > packets like it should, and the cisco router sees them, and show ip wccp
> > web-cache detail shows them as usable, there's just no hash assigned to
> > them (and when the designated cache reconfigures the hash without any
> > change notification, the cache gets his share)
> >
> > Btw when I searched the web about this problem, I came to a page where
> > they wrote about a similar problem with the cisco cache engine. The
> > solution (hah!) explained there was to disable wccp for 30 seconds, then
> > enable it again on the cache engine. So it might be a problem with the
> > wccp protocol after all...
>
> Well, wccp2 most certainly has that in the protocol spec. I haven't
> read the wccpv1 docs. As for wccp2 .. if you really need it .. :)
>
>
>
>
> Adrian
Received on Fri Feb 15 2002 - 06:04:44 MST