On 12/05/06, sebb <[EMAIL PROTECTED]> wrote:
On 12/05/06, [EMAIL PROTECTED]
<[EMAIL PROTECTED]> wrote:
>
> Thanks Milla and Durgs for your information sharing.
>
> But we encountered this problem with even 30 users only.
>
> And as per our scenarios, 30 users are not considered as load over the
> application as we need to cover much higher than that.
>
> Due to this we said that the load testing is getting affected and nearly
> stopped.
JMeter can hopefully be changed to stop the error in one thread from
bringing down the whole test. I think I have identified a bug whereby
the streams are not being properly closed when an error occurs while
reading a response in the default HTTP Sampler.
However, there's not a lot JMeter can do to stop the error from
occurring in the first place - ultimately, that is caused by the
server under test.
All you can do at present is to try with Http Client instead, and see
if that helps.
[The bug I mentioned is for the default sampler only].
==
I have tried testing with a JSP that waits for a configurable length
of time, but could not reproduce the error even when the wait was 6
minutes. So if there is a server I could use to test against, please
let me know (privately if necessary).
==
I do want to try and fix what can be fixed, but it's going to take some time:
- SVN is down currently
Now fixed.
- there is no test case, so I'm having to guess what might help
Still true.
- other commitments
I've applied a fix to the code, which is in the nightly version 2-1.20060514.
Hopefully this will stop the reported error from stopping all the
threads, but I have no means of testing this, so I don't know if the
fix will work or not.
Please try the new release and report if it solves the problem or not.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
