Mailing-List: contact tomcat-dev-help@jakarta.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Tomcat Developers List" <tomcat-dev@jakarta.apache.org>
To: tomcat-dev@jakarta.apache.org
From: "Bill Barker" <wbarker@wilshire.com>
Subject: Re: LB with mod_jk 2
Date: Fri, 21 Nov 2003 22:00:02 -0800
Lines: 27
Message-ID: <bpmtbc$cvj$1@sea.gmane.org>
References: <LAW10-F52YqHSPVYAtj00002605@hotmail.com>
 <3FBE373B.1040207@tagish.com> <3FBE3D67.4090200@apache.org>
Sender: news <news@sea.gmane.org>


"Remy Maucherat" <remm@apache.org> wrote in message
news:3FBE3D67.4090200@apache.org...
> Hi,
>
> I got a support question about LB with mod_jk 2 (and I saw posts about
> that on tomcat-user).
>
> Either there's something I don't understand, or the
> jk2_get_most_suitable_worker method in jk_worker_lb will always return
> the first worker which is not in an error state (assuming there was no
> session), causing all requests to be redirected to the first declared
> worker in a group.
>
> So did I miss something, or is the algorithm bad ?
>

My recollection is that the algorithm is bad.  Of course, I'm currently of
the opinion that mod_jk2 is hopelessly broken, so I don't spend much time
there :).  At best, it suffers from the same problems as mod_jk if you are
using the pre-fork MPM, so it doesn't work at all.  mod_jk2 will need to
save this to the smem file in order to work with pre-fork.

> Thanks,
> R�my


---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org