You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jmeter-dev@jakarta.apache.org by Oleg Kalnichevski <ol...@apache.org> on 2007/07/05 16:12:27 UTC
Re: Re: Performance issue =?ISO-8859-1?Q?s with 2.3RC1?=
On Thu, 2007-07-05 at 16:05 +0200, jens@voss-ahrensburg.de wrote:
> Okay, it looks like I isolated the issue: There seems to be a problem
> with the SSL manager: Some of the pages we load in the test plans
> contain framesets with frames pointing to https addresses served by our
> proxy. In this scenario the log file shows a NullPointerException (and
> the single HTTP sampler (including embedded page loading) takes up to
> three minutes to execute).
>
> I have submitted this as bugzilla bug #42819.
>
> Best regards,
> Jens
>
>
Jens,
Are there multiple threads involved in executing that sample? If there
are, it can easily explain the performance degradation you have been
seeing. JMeter now creates a separate SSL context per thread of
execution, which is a _very_ expensive operation. For details see bug
#42506 [1]
Oleg
[1] http://issues.apache.org/bugzilla/show_bug.cgi?id=42506
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jmeter-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: jmeter-dev-help@jakarta.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-dev-help@jakarta.apache.org
Re: Performance issues with 2.3RC1
Posted by sebb <se...@gmail.com>.
On 12/07/07, Jens Voß <je...@voss-ahrensburg.de> wrote:
> sebb wrote:
>
> > Just to follow up - can I just confirm that the performance problems
> > have gone away - now that the issue with the password / NPE has been
> > sorted out?
> >
> > Seems to be the case from the Bugzilla issue, but it would be useful
> > to get it confirmed here (if you don't mind).
>
> Yes, no more performance problems; the test plans in our suite now
> run about 1.5 times faster with 2.3RC1 than with 2.2 on average.
Thanks, that's excellent news.
BTW, RC3 should be the same (only changes were to fix the NPE you
reported, some test localisation errors and some documentation
tidyups).
> Best regards,
> Jens
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jmeter-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: jmeter-dev-help@jakarta.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-dev-help@jakarta.apache.org
Re: Performance issues with 2.3RC1
Posted by Jens Voß <je...@voss-ahrensburg.de>.
sebb wrote:
> Just to follow up - can I just confirm that the performance problems
> have gone away - now that the issue with the password / NPE has been
> sorted out?
>
> Seems to be the case from the Bugzilla issue, but it would be useful
> to get it confirmed here (if you don't mind).
Yes, no more performance problems; the test plans in our suite now
run about 1.5 times faster with 2.3RC1 than with 2.2 on average.
Best regards,
Jens
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-dev-help@jakarta.apache.org
Re: Re: Performance issues with 2.3RC1
Posted by sebb <se...@gmail.com>.
Just to follow up - can I just confirm that the performance problems
have gone away - now that the issue with the password / NPE has been
sorted out?
Seems to be the case from the Bugzilla issue, but it would be useful
to get it confirmed here (if you don't mind).
Thanks,
S
On 05/07/07, sebb <se...@gmail.com> wrote:
> On 05/07/07, Oleg Kalnichevski <ol...@apache.org> wrote:
> > On Thu, 2007-07-05 at 16:05 +0200, jens@voss-ahrensburg.de wrote:
> > > Okay, it looks like I isolated the issue: There seems to be a problem
> > > with the SSL manager: Some of the pages we load in the test plans
> > > contain framesets with frames pointing to https addresses served by our
> > > proxy. In this scenario the log file shows a NullPointerException (and
> > > the single HTTP sampler (including embedded page loading) takes up to
> > > three minutes to execute).
> > >
> > > I have submitted this as bugzilla bug #42819.
>
> OK, I'll look at that ASAP.
>
> > >
> > > Best regards,
> > > Jens
> > >
> > >
> >
> > Jens,
> >
> > Are there multiple threads involved in executing that sample? If there
> > are, it can easily explain the performance degradation you have been
> > seeing. JMeter now creates a separate SSL context per thread of
> > execution, which is a _very_ expensive operation. For details see bug
> > #42506 [1]
> >
> > Oleg
> >
> > [1] http://issues.apache.org/bugzilla/show_bug.cgi?id=42506
> >
>
> You can try defining the jmeter property:
>
> https.sessioncontext.shared=true
>
> to revert to the previous behaviour.
>
> What happens if you try that?
>
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-dev-help@jakarta.apache.org
Re: Re: Performance issues with 2.3RC1
Posted by sebb <se...@gmail.com>.
On 05/07/07, Oleg Kalnichevski <ol...@apache.org> wrote:
> On Thu, 2007-07-05 at 16:05 +0200, jens@voss-ahrensburg.de wrote:
> > Okay, it looks like I isolated the issue: There seems to be a problem
> > with the SSL manager: Some of the pages we load in the test plans
> > contain framesets with frames pointing to https addresses served by our
> > proxy. In this scenario the log file shows a NullPointerException (and
> > the single HTTP sampler (including embedded page loading) takes up to
> > three minutes to execute).
> >
> > I have submitted this as bugzilla bug #42819.
OK, I'll look at that ASAP.
> >
> > Best regards,
> > Jens
> >
> >
>
> Jens,
>
> Are there multiple threads involved in executing that sample? If there
> are, it can easily explain the performance degradation you have been
> seeing. JMeter now creates a separate SSL context per thread of
> execution, which is a _very_ expensive operation. For details see bug
> #42506 [1]
>
> Oleg
>
> [1] http://issues.apache.org/bugzilla/show_bug.cgi?id=42506
>
You can try defining the jmeter property:
https.sessioncontext.shared=true
to revert to the previous behaviour.
What happens if you try that?
---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-dev-help@jakarta.apache.org