tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kristian Rink <>
Subject clustering and jvmRoute
Date Fri, 23 Feb 2007 07:50:01 GMT


yesterday I dealt with the tomcat clustering feature for the first
time, getting things basically set up and working in quite a short time
(thanks to the documentation which is rather good), and even made my
application distributable / the session attributes serializable really
fast. However, one thing is massively confusing me:

I followed the straightforward approach, reading through [1] where
there is a configuration file in a section named "Simple Engine Cluster
Configuration for all hosts" - without thinking twice, I installed a
configuration similar to that on both of the nodes of my cluster
(including the _same_ jvmRoute parameter on both machines) and it
worked - no matter which one of the nodes went down, my session
remained alive and well.

However, spending some more time trying to study the details and
options of clustering, I came across [2] and learnt that the jvmRoute
parameter should be a per-host not a per-cluster configuration.
Thinking twice about this and its implications on session replication,
this seems reasonable, and looking at the log of mod_jk while running
both tomcats with the same jvmRoute parameter, indeed I see requests
belonging to the same session processed by different nodes (which, if I
got the servlet specifications right, shouldn't be). So: Why does this
setup, having jvmRoute the same in all cluster nodes, work? Just an
undesired side-effect of a broken configuration? Luck? Is there
something I didn't yet get about the cluster configuration?

Thanks in ad


Kristian Rink * *
jab: * icq: 48874445 * fon: ++49 176 2447 2771
"One dreaming alone, it will be only a dream; many dreaming together
is the beginning of a new reality." (Hundertwasser)

To start a new topic, e-mail:
To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message