On 26/11/2008, Steve Kapinos <[EMAIL PROTECTED]> wrote: > One of my recent runs of a nightly simultation melted down on a certain > run. Trying to piece together why. The jmeter log is full of > exceptions like this > > 2008/11/25 13:48:19 INFO - jmeter.control.IncludeController: > loadIncludedElements -- try to load included module: > vcs-calllog-generator.jmx > 2008/11/25 13:48:19 INFO - jmeter.engine.StandardJMeterEngine: Starting > 40 threads for group VCS Thread Group. Ramp up = 10. > 2008/11/25 13:48:19 INFO - jmeter.engine.StandardJMeterEngine: Continue > on error > 2008/11/25 13:48:19 INFO - > org.apache.commons.httpclient.HttpMethodDirector: I/O exception > (java.net.BindException) caught when processing request: Address already > in use: connect > 2008/11/25 13:48:19 INFO - > org.apache.commons.httpclient.HttpMethodDirector: Retrying request > 2008/11/25 13:48:19 INFO - > org.apache.commons.httpclient.HttpMethodDirector: I/O exception > (java.net.BindException) caught when processing request: Address already > in use: connect > 2008/11/25 13:48:19 INFO - > org.apache.commons.httpclient.HttpMethodDirector: Retrying request > 2008/11/25 13:48:19 INFO - > org.apache.commons.httpclient.HttpMethodDirector: I/O exception > (java.net.BindException) caught when processing request: Address already > in use: connect > 2008/11/25 13:48:19 INFO - > org.apache.commons.httpclient.HttpMethodDirector: Retrying request > > And eventually you see the retries failing > > 2008/11/25 13:48:26 ERROR - jmeter.protocol.http.sampler.HTTPSampler: > Can't connect after 10 retries, java.net.BindException: Address already > in use: connect
Are you sure you copied that last log entry correctly? Or have you got both types of HTTP Sampler in the test plan? The last log message is only used by the original (Java) sampler, and the others are all the HttpClient version. > Is this java complaining it can not bind to a local socket to open the > http sampler? > Lack of sockets, probably. I think the error message is a bit misleading - it should probably be something more like "socket not available". As far as I can tell, client sockets call bind() when establishing the ephemeral socket - this results in a bind error rather than a create/open error. > Any idea what would cause java to melt down like this? Make sure you are using keep-alive where necessary to reduce connection churn. > It's a Win2003 server that normally runs the test w/o issue and even did > the following nights Maybe the server sometimes uses additional sockets. > -Steve > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
