[jboss-user] [Clustering] - client takes long to fail over to other node

[justkeys]

I read about jnp.timeout and jnp.sotimeout here  http://www.jboss.org/community/wiki/NamingContextFactory but these timeouts seem only to affect the lookup of the Naming object; this does not seem to affect the time it takes, for a proxy to decide that node A is no longer reachable, and node B should be tried.

Say i have a 2-node cluster, i lookup "MyService/remote" (using the HA context :1100) and store this proxy for later use. Then I remove one node from the cluster (disconnect it from the network): the proxy should be aware of 2 server nodes (since both were up when i created it)  and it should try them both, before failing. This often takes much longer than my configured timeouts. Is there any way to influence this: configure the socket-timeout of a communication-attempt for a particular proxy?

View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4257948#4257948

Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4257948