On 08.06.2009 13:22, Darren Kukulka wrote: > Hi Everyone, > > > > I've got a general query around application clustering in Tomcat. > > > > We've got several clusters of Tomcat 6.0.18, each consisting of 2-3 > nodes per cluster. The setup is fairly straightforward and we're using > the 'based on load' balance method at the web server layer for the ajp > workers. > > > > This works great for scenarios where cluster members are bought in and > out of the cluster groups, e.g. when bouncing Tomcat. > > > > My question is whether the cluster engine can be configured to failover > to a working node when an application becomes unavailable on another > node, i.e. when an application is bought down, rather than the Tomcat > instance, using Lambda Probe for example. > > > > Can anybody please point me in the right direction?
If you are using mod_jk (?), then you can use the so-called status worker to let mod_jk know, that you do no longer forward traffix to some nodes. This can also be scripted. If you only want to stop forwarding for some JkMounts, then look at the possibilities of uriworkermap.properties, and the so-called mount extensions. They allow to dynamically stop forwarding of individual mounts to some members of a load balancing worker. Automatic detection of an unavailable context is not very robust. You might have a look at the worker attribute fail_on_status. Regards, Rainer --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
