I was mistaken - what happens is that I recieved a warning notice "service did not start within 120sec" at the same time as the microcontainer died. I don't know where this message comes from - it is a gui popup, so probably something in JBoss IDE.

I figured out that microcontainer must have run out of memory. It just happened to occur at the exact same time as the timeout message appeared. No decent error message though (started from within JBoss IDE) - it just says "deployment failed" or something like that - no details.

After having increased the memory size I was able to debug.

I finally got this problem solved, and want to share the solution.

After upgrade to JBoss 4.2.1.GA I noticied that there was a warning message that a port that was wanted was used by someone else. The app server still started.

A collegue pointed out, that when Skype or IP phone software (and some chat software) is started before jBoss, they tend to occupy ports that are wanted by JBoss by default.

So, starting debugging before starting Skype (and in my case Trillian) reduced the startup time from over 10 minutes to sub minute!!!

Seems a little bit to strange to be true. What do you think? Could this have been what was causing my problems?