Hi,
Running Apache 1.3.14, Win 2k pro, and tomcat 3.2.1, with mod_jk, and with pretty much the same setup, only with 2 tomcats, it did reconnect to the failed server after about 1 minute.
jk_lb_worker.c says that:
#define WAIT_BEFORE_RECOVER (60*1)
which is the time in seconds before mod_jk tries to reconnect a failed tomcat. Fancy that. However, it is the minimum time it will wait. If you have no incoming requests, it will not reconnect. I think. See below.
You asked: "Once server C comes back on line, do we have to do anything specific to
inform the load balancers or will they try to connect to a failed server/work on a ragular basis"
From looking at jk_lb_worker in tomcat/src/native/jk, it seems to me that every time there is a request, it tries to get the most suitable worker. If there is a worker that has failed, but has failed more than WAIT_BEFORE_RECOVER seconds ago, it will mark it as recovering, and return it as a worker. If it is still failed then it is again marked as failed, and it tries to find another worker.
So, the answer to your question is no, yes, and but. :) No, you do not have to do anything specific to inform the load balancer that a tomcat has recovered. Yes, they will try to connect to a failed server on a regular basis. But, there have to be incoming requests to apache for it to try to connect to a failed server.
Hope this helps,
Jord.
-----Original Message-----
From: Steven Toth [mailto:[EMAIL PROTECTED]]
Sent: Monday, January 29, 2001 11:38 AM
To: [EMAIL PROTECTED]
Subject: mod_jk load balancing error mechanisms during tomcat failure
Hi,
A question ... but before I ask ... I'll give everyone a little
background...
I'm working on an application that we want to be able to scale n-way with a
number of apache servers (using mod_jk and load balancing) to a number of
tomcat servers running on seperate dedicated servers.
In practise, server A (running apache and mod_jk) is configured to load
balance between servers C,D,E and F. Server B (also running apache and
mod_jk with load balancing) will balance between the same set of servers
C,D,E anf F.
We're also using the DNS round-robin for distribute the incoming http
requests between servers A amd B.
My question is... Assuming server C dies. The load balancers on servers A
and B will eventually notice and redirect the requests to either D,E or F.
That's fine and it seems to work.
Once server C comes back on line, do we have to do anything specific to
inform the load balancers or will they try to connect to a failed
server/work on a ragular basis
I've tested a little and the results indicate that it doesn't notice a
failed servers resuming operation. Anyone know for sure what happens?
Steve
_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]
