> What kind of load are you doing?
i loadtest with the openwebload tool (
http://openwebload.sourceforge.net) which simulates concurrent access of different clients.
> Can you do a netstat -an | grep 8080 (assuming 8080 is what you are using).
tcp 0 0 192.168.6.69:8080 192.168.6.66:50964 TIME_WAIT
tcp 118 0 192.168.6.69:8080 192.168.6.66:51116 VERBUNDEN
tcp 0 0 192.168.6.69:8080 192.168.6.66:50117 TIME_WAIT
tcp 118 0 192.168.6.69:8080 192.168.6.66:51205 VERBUNDEN
tcp 118 0 192.168.6.69:8080 192.168.6.66:51145 VERBUNDEN
tcp 0 0 192.168.6.69:8080 192.168.6.66:50218 TIME_WAIT
tcp 118 0 192.168.6.69:8080 192.168.6.66:51090 VERBUNDEN
tcp 118 0 192.168.6.69:8080 192.168.6.66:51092 VERBUNDEN
tcp 0 0 192.168.6.69:8080 192.168.6.66:50309 TIME_WAIT
VERBUNDEN = CONNECTED.
> If you are doing keep-alived connections, flooding the server will produce such
> exception as the
> connection stay opens for 30 seconds. If you expect
> such 'flooding'
> traffic, you might want to throttles requests by
> reducing the idle time
> from 30 seconds to something lower.
>Which clients are
> you using to
> stress the server?
yesterday i still changed the "Commit Option" of the EJB-Container in the web-admin-console from B to C and the error does not come sooo early anymore.. but at higher rates the server still dies with this error.
so the client is just basically the openwebload tool which calls a jsp which itself calls a SLSB. should not be the big deal. but maybe you can also take a look at my related thread here:
http://forums.java.net/jive/thread.jspa?threadID=37132&tstart=0
thanks for helping,
greetings,
mario
[Message sent by forum member 'the666pack' (the666pack)]
http://forums.java.net/jive/thread.jspa?messageID=260831