Mike Finn wrote: > I had been working on an alternative solution to this problem (see earlier thread in >Mid-April: "Multiple Instances"). After some thought and work, this (Dain's) seems >more robust (at least to me). The other way was effectively a proxy server sitting on >a single (known) port, directing streams to the correct endpoints. The problem is >that not every wire protocol puts enough (or anything) in the initial packets in the >stream to identify the service - so I don't think the proxy would work as we >discussed before (HTTP would work OK, but it looks like JNDI and RMI wouldn't). >Unfortunately, I didn't do any packet sniffing to find this until I had already >written the proxy server/MBean :-(. > > Q > 1) I assume "service name" would be the JBoss service name? (jmx-html, webserver, >etc). IP service name would be insufficient as there are multiple services serving >HTTP (webserver, jmx-html, jetty/tomcat). > > 2) How do clients find what port to which they connect today? For example, w/ a >separate-VM-client that does a bean lookup (via configured port - eg 1099) - where >does it find the RMI port number(eg 4444) to make the method call? Is it in the >object returned from the JNDI server? (Probably a dumb one, but not obvious to me).
I've been assuming that they'd be statically mapped - There are a few things (JNDI and servlet-HTTP at leaast) that have to be consistent from run to run for a particular server config. At that point, a standalone client would mostly need its initial context URL overriden - each developer just keeps their own jndi.properties for such a client? -danch _______________________________________________________________ Don't miss the 2002 Sprint PCS Application Developer's Conference August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm _______________________________________________ Jboss-development mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-development
