Jordi and Laurent -
Thanks, I've updated the bug with my observations. I confirm the bug on
1.9.1 and CVS.
I don't know my way around SSL/PK in Java (I used openssl/Unix etc). I'd
be grateful if you told me which debug settings you use. to get those
stacktraces. I tried the following switches (saw them in the Developer
manual): -Ljmeter.util=DEBUG , -Ljmeter.engine=DEBUG , -LDEBUG .
However, no information was output - either to stdout or to a disk file.
I like comments on my approach to surge testing. Is it valid?
Also, is there an equivalent of a SimpleController, but for Threads?
This could be used like this:
---------------------------------------------------------
Test Plan -- Runs thread groups one after another
Dummy Group 1 -- Dummy SSL connection group
>>>> Simple Thread Holder <<<< Runs thread groups *concurrently*
Thread Group 2 -- Normal usage group
Thread Group 3 -- Surge group.
---------------------------------------------------------
Group 1 would run first, then Group 2 and 3 would run together.
With regards,
Sonam Chauhan
--
Corporate Express Australia Ltd.
Phone: +61-2-9335-0725, Fax: 9335-0753, Email: [EMAIL PROTECTED]
> -----Original Message-----
> From: Jordi Salvat i Alabart [mailto:[EMAIL PROTECTED]
> Sent: Wednesday, 17 December 2003 9:39 PM
> To: JMeter Users List
> Subject: Re: surge testing and SSL connection bug
>
> Hi Sonam,
>
> I'll be grateful if you post whatever you learn about this issue into
> the bug report -- I've not even researched it yet, so Laurent's
baseline
> that we're missing some configuration step may well hold.
>
> --
> Salut,
>
> Jordi.
>
> En/na Laurent MEDIONI ha escrit:
> > Hi,
> > I have never experienced Jordi's bug
> > (http://nagoya.apache.org/bugzilla/show_bug.cgi?id=25505) using the
same
> jdk
> > and 1.9.1.
> > Have you tried to increase the log level for both https and jmeter ?
> > You may have a problem to open your keystore ? How do you specify
the
> > keystore path, name and password ? In the JMeter conf file ? Do you
have
> the
> > necessary certificate and CA in this keystore ? Do you use the
> sslManager to
> > import your cert mag?
> >
> > Be careful that surge testing https will (artificialy) over-stress
your
> ssl
> > gateway because of the non-reuse of the connections by each thread
> > (http://nagoya.apache.org/bugzilla/show_bug.cgi?id=22512). And never
use
> the
> > "keep alive" option on the samplers because you would soon also
reach
> the
> > limit of your local https connection stack (Windows issue).
> >
> > Cheers,
> > Laurent
> >
> > -----Message d'origine-----
> > De : Sonam Chauhan [mailto:[EMAIL PROTECTED]
> > Envoy� : mercredi, 17. d�cembre 2003 08:11
> > � : 'JMeter Users List'
> > Objet : surge testing and SSL connection bug
> >
> >
> > Hello JMeter users/developers-
> >
> > I found an SSL bug in the latest JMeter where it throws an exception
on
> the
> > very first SSL connection. I saw Jordi had already filed it in
bugzilla:
> > http://nagoya.apache.org/bugzilla/show_bug.cgi?id=25505
> > Jordi states a workaround:
> > "[Workaround: create a stupid sample at the very beginning of
your
> > test]"
> > This works a treat. However, it breaks something I am planning to
do,
> and I
> > was wondering if there was a better way...
> >
> > I want to implement a "surge test" where the number of users
"surges"
> > temporarily every hour or so, then dies down quickly. So my plan has
two
> > thread groups like so:
> > ---------------------------------------------------------
> > Test Plan -- Runs thread groups *simultaneously*
> > Thread Group 1 -- Normal usage group
> > Thread Group 2 -- Surge group.
> > ---------------------------------------------------------
> > The surge group finishes one iteration in a few minutes. A constant
> delay
> > timer at the end causes it to sleep for an hour before running
again.
> >
> > The problem is to implement the workaround for HTTPS, I will have to
> insert
> > a dummy first thread group. I will also have to tick the "Run each
> > ---------------------------------------------------------
> > Test Plan -- Runs thread groups *one after another*
> > Thread Group 1 -- *Dummy SSL connection group*
> > Thread Group 2 -- Normal usage group
> > Thread Group 2 -- Surge group.
> > ---------------------------------------------------------
> > To ensure dummy Thread Group 1 runs first, I have to tick "Run each
> thread
> > groups separately" in the top level plan element. But this breaks
the
> surge
> > testing because thread groups 2 and 3 no longer run together.
> >
> > Is there a better way to do surge testing?
> >
> > With regards,
> > Sonam Chauhan
>
>
> ---------------------------------------------------------------------
> 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]
