The HTTP threads settings are in "$JBOSS_HOME/server/default/deploy/jbossweb.sar/server.xml". Look at the documention for Tomcat to see which settings you want to change in the "http connector"
Thanks for the reply. That is where I used to set the maxThreads, minSpareThreads, maxSpareThreads and acceptCount for the connector in JBoss 4.2.3.
The format of the file is slightly different for AS 6 and the these properties are missing. The jmx-console allows me to set these values.
I added these properties to the server.xml and server started up fine, with my new values.
My only concern is, is that the right way to set values?
What about thread-pool-jboss-beans.xml file? How is that used?
Here you might find hekpful links (maybe later)
also I think the Tuning and sliming of previous versions will not completely outdated and helpful.
That is where I used to set the maxThreads, minSpareThreads, maxSpareThreads and acceptCount for the connector in JBoss 4.2.3.
Yes, but those have not been the way to configure HTTP threads since Tomcat 6 (or since JBoss AS 5.0.0). See this:
I didn't know about the executor. If I understand it correctly the "old" setup is valid if you are not using the
executor settings in the connector to point out the thread pool to use.
My experience has been that the "old" values are ignored. I recall some discussions about this a year or two back, I think that the poster was wondering why he had more threads than maxThreads (or perhaps why on an idle system the thread count did not go down to minSpareThreads). If I find the thread I'll post the link.
Yes, agreed. But the real question is - do I still modify the server.xml to configure tomcat?
What I am seeing it that under load (using JMeter, when I fire 300 threads, with no delay), JBoss puases.
100/200 Threads, no problem.
What am I doing wrong?
do I still modify the server.xml to configure tomcat?
What am I doing wrong?
Could be any number of things. To begin with, I would not fire off 300 threads at once. There are way too many things that can go wrong. I would place a delay of about 100-200 milliseconds between each thread, minimum. But without examining your test environment and taking a whole lot of measurements, there is no way to tell what the problem is.