29378 is not fixable, as far as I can make out. There is no control
over connection re-use with the default Java HTTP implementation. I've
closed the bug.

31832 may perhaps be caused by the default Java implementation running
out of connections? It may or may not be fixable. Perhaps you could
use netstat or similar to see how many connections are open when the
process hangs. Are you using "Download Embedded resources"?

I've made quite a few improvements to the code just recently, and will
generate a new "nightly" build today or tomorrow.

This won't help with the default HTTP sampler, but the HTTPClient
Sampler has been updated quite a bit, and should hopefully work
better. If not, then at least there is a chance of fixing it.

If you possibly can, please try using the HTTPClient HTTP Sampler -
it's easy enough to edit the JMX file if you don't want to recreate
it. See the file extras/ConvertHTTPSampler.txt for details.

Hopefully this will help you - it would certainly help us if the new
code was given a full work-out!

[I'm working on making it possible to change the HTTP implementation
in the GUI - nearly there...]

S.
On 30/04/06, Oliver Erlewein (DSLWN) <[EMAIL PROTECTED]> wrote:
Hi Sebb,

After a LOT of testing and trying I think what I'm seeing are the
problems described in Bug 31832 and Bug 29378. Do you have an idea when
those bugs will be fixed or whether there's some kind of work around?

Oliver

-----Original Message-----
From: sebb [mailto:[EMAIL PROTECTED]
Sent: Tuesday, 18 April 2006 21:43
To: JMeter Users List
Subject: Re: Jmeter problems 2h into perf test

We often run tests over a weekend, and have not seen this behaviour.
The test runs can use 100,000 samples or more.

However, we only use non-GUI mode, and we use only the listener created
by the -l flag, writing to a CSV file. [Plus a Summariser, so we can
keep track of the test progress.]

Make sure you are not using any unnecessary listeners - the aggregate
listener (and some others) currently keeps a copy of every sample...

S.

---------------------------------------------------------------------
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]

Reply via email to