At 16:29 +0200 24-09-2004, Gerard van Enk wrote:
http://www.mmbase.org/testing/?parameters=portal,page&portal=199&page=546&btemplate=fullview.jsp&bugreport=25953
Ah, ok, I still don't know whether this is a real bug in MMBase or a configuration problem.
I'm sorry?!
If it is a real bug, why is not everybody having the same problems? The vpro doesn't have this problem, the eo doesn't have this problem. I'm not stating it *isn't* an MMBase bug, but I'm not sure.
At 22:53 +0200 22-09-2004, Nico Klasens wrote:
The multicast implementation already used 2 queues for sending and
receiving. The receiving queue was read by a thread which spawned for every
message a new thread.
If this is true what Nico found out and the receiving end really spawns a thread for every message it receives, it surely is a bug.
Why does this have to be a bug?
I've seen a JVM running MMBase 1.7 crash at 820 Mb with a java.lang.OutOfMemory error while it had 1,5 Gb allocated.
I've seen a lot of jvm crashes and most of them were cause by misconfiguration or by bugs in the JVM. That doesn't mean MMBase can't be the cause of memory problems.
The problem you're describing above sounds like one of the memory spaces inside the jvm is full, so this *could* mean it's a configuration problem.
Gerard
