Home.... | log in.... | Pub my question! | Sign-up....

JVM in RMID vs Standalone JVM

Isn't it the the mechanism/process/memory model in standalone JVM same with the JVM spawned by the RMI Activation server?

This is regarding the creation of multiple native objects (C++). It works fine with a standalone program written in Java through JNI. However, the JVM hanged during the creation of native objects when it was running within a RMI server. No stack traces printed, the RMI server then cannot be stopped by rmid -stop, though it seemed like stopped sucessfully, but all the java and rmid processes were still exist and have to be terminated by kill command under Linux.

[601 byte] By [tycuzza] at [2007-10-2 22:01:14]

«« JSTL is this correct?
»» JSP with XML SOAP

# 1

I'm not an RMID expert, but I think the difference is in the fact that RMID will create a new JVM for each request.

If you use a single JVM for all your remote objects, there may be some low level interaction between different native objects (e.g, global variables, memory allocation/access errors) that you wouldn't otherwise encounter.

genadya at 2007-7-14 1:17:30 >

# 2

No. RMID creates a new JVM per ActivationGroup, which is the thing that the Activatable was registered inside when it was registered. When any activatable in that activation group has to be activated a JVM is created for the group if it isn't already running.

ejpa at 2007-7-14 1:17:31 >

# 3

Sounds like a better idea :)Thanks,Genady

genadya at 2007-7-14 1:17:31 >

# 4

Problem has been found, there was a deadlock int the C++ program. Everything works fine now.But I still don't understand why I couldn't reproduce the problem under a stand-alone Java program.Thanks anyway!

tycuzza at 2007-7-14 1:17:31 >

Java programming resources: FAQs, tutorials, compiler and browser download sites, documentation, books lists, IDEs, etc.

JVM in RMID vs Standalone JVM

Core New Topic

Core Hot Topic

Core

All Classified