I just checked with the design guys and it seems we are using cookies
after all, by default (sorry Peter for not
checking this earlier). I will use Cookie Manager for the request and this
should fix the problem,
Thanks again,
Alan
sebb <[EMAIL PROTECTED]>
04/04/2005 12:53 PM
Please respond to "JMeter Users List"
To: JMeter Users List <[email protected]>, Peter Lin
<[EMAIL PROTECTED]>
cc:
Subject: Re: Giving Http requests access to HttpSession in
JMeter tests
On Apr 4, 2005 12:44 PM, Peter Lin <[EMAIL PROTECTED]> wrote:
> On Apr 4, 2005 4:23 AM, [EMAIL PROTECTED] <[EMAIL PROTECTED]>
wrote:
> > Thanks Sebb,
> >
> > I found out why I am having the problem. The first struts call I make
to
> > the application uses a redirect
> > depending on the input (if only one input entry, build 1st page and
then
> > goto 2nd page, otherwise go to first page).
> > With my example input, there was always a redirect to a second ".do"
and
> > the session was being lost. It works ok
> > if I use input that doesn't cause a redirect and only calls a single
> > ".do".
> > I cannot use the cookie manager as I am not using cookies in the
> > application - sesison data is maintained by
> > the server. Is there any reason why JMeter loses the Http Session on a
> > redirect and would there be any way
> > to resolve this?
>
> I'm not sure I understand this. If you're not using cookies to store
> session state, that means it is using request parameters. If it is
> using request parameters for session id, like the way Yahoo does it,
> then you will need to get the correct req param and pass it back to
> the server.
>
Indeed. Is a browser guaranteed to maintain a connection to a server?
Seems to me that if the application is relying on the client reusing
the same connection, it may fail with some clients - or with some
clients some of the time.
There's no control over connection re-use with the default HTTP
implementation.
The Apache HttpClient version does try to re-use connections, but at
present only for the same sampler in the same thread (this is a bug).
Having said this, it might be worth trying "redirect automatically" as
that is done by the underlying HTTP implementation, and it may well
use the same connection - if that's what you really need.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
