Are you intentionally deploying an EJB app on start-up? It certainly looks like
an EJB or the EJB service is failing to load, but I don't know if it's one of
yours or one of ours (like a sample you deployed or something). Have you always
had this problem or is this only since you've tried to deploy something? If the
latter, I would clean out your deploy and runtime directories, and even disable
the ejb service in global.properties until we figure it out.
As for the zombie threads, is top telling you that the threads are zombies,
i.e., process in state Z? If so, that's unusual and I'm not sure what it has to
do with this problem. But like I said, the zombies themselves don't use memory.
You must be killing their parent when you do the killall.
We certainly have JRun 3.0 running on plenty of Linux machines at work, so it
makes me wonder if it's something you tried to deploy or a configuration change
you made, or if you're having this problem "out of the box." How are you
starting and stopping JRun? Just using the jrun command? With which option(s)?
Scott Stirling
-----Original Message-----
From: Matt Senter [mailto:[EMAIL PROTECTED]]
Sent: Monday, July 10, 2000 3:21 AM
To: [EMAIL PROTECTED]
Subject: RE: Zombie threads
Well, when I completely stop JRun, I can run "top" and sort by memory usage,
and a lot of Java threads appear first in line. Plus, it says there are
anywhere in the area of 18 zombies, so I was assuming those threads were the
zombies. When I "killall java" and run "top" again, there are no more
zombies...except for a few pesky lingering java threads that just won't
die...plus after I "killall" my free memory skyrockets. You'll have to
forgive me, i'm not the most knowledgable Linux user, so maybe there's
something else still running from JRun that's taking up that memory, but I
just figured it was those java threads because the memory wouldn't free
until I killed them. I checked all the ports, and there aren't any
conflicts, and jrun never gives me a port conflict error. Any other ideas? I
really think it has something to do with this error i get on loading:
Operating System: Linux Version 2.2.14-6.0.1
Java Virtual Machine: build Linux_JDK_1.2.2_RC4, native threads, sunwjit fr
om Blackdown Java-Linux Team
JRun 3.0 3.00.3694 Starting default...
Current Locale: en_US
Loading scheduler
Loading logging
Exception: [10:36:43] Unable to install java:comp/env/ejb/TxnHome
Exception: [10:36:43] Unable to install default.RoleHome
Exception: [10:36:44] Unable to install default.QueueConnectorHome
Exception: [10:36:45] Unable to install default.LoginSessionHome
Exception: [10:36:45] Unable to install default.TopicConnectorHome
Exception: [10:36:45] Unable to install default.UserHome
Exception: [10:36:46] java.lang.IllegalStateException: No default.UserHome
found
Thanks,
-Matt
-----Original Message-----
From: Scott Stirling [mailto:[EMAIL PROTECTED]]
Sent: Saturday, July 08, 2000 9:00 AM
To: [EMAIL PROTECTED]
Subject: RE: Zombie threads
One issue seems to be that your default server configuration is messed up. I
would look for port conflicts with the EJB server or something. Check your
local.properties for all your servers and make sure none of the ports are
the
same (a tool for this would be helpful, wouldn't it?).
For the parent of the "zombie" processes, I would be interested in a stack
trace. How do you know these threads are zombies?
Note - zombie processes don't "eat up" memory. The only resource they
consume
is the space for an entry in the process table. If you kill the parent
process
of a zombie, then the zombie should disappear from ps.
Scott Stirling
------------------------------------------------------------------------------
Archives: http://www.egroups.com/group/jrun-interest/
Unsubscribe: http://www.houseoffusion.com/index.cfm?sidebar=lists&body=lists/jrun_talk
or send a message to [EMAIL PROTECTED] with 'unsubscribe' in the
body.
