Slow respone of IdP in Jmeter tests

[Peter Schober]

Some more very minor comments which will not help with yuor Jmeter
setup, I'm afraid.

* Gernot Hassenpflug <ha4h-grnt at asahi-net.or.jp> [2013-01-18 06:06]:
> Sorry, I don't know where that came from. We are using IdP version
> 2.3.6. Previously we used 2.1.5, which had Tomcat5 as a
> recommendation. I did not read the documentation for 2.3.6 carefully
> enough to realize that Tomcat6 is required, oops!

IIRC Tomcat < 6.0.17 had an issue with cookies as used by the IdP --
or the other way round -- affecting the IdP version that was
introduced around August 2009, from a quick look.
But the way you exercise the IdP during loadtesting probably doesn't
even make any use of those cookies, hence Tomcat5 would "work".

> > Also the documentation states Tomcat 6.0.17 or higher 6.x as
> > a requirement.
> 
> I will try again with 2.1.5 and see if there is a difference.

Based on the above (and 2.1.5 being historic) I wouldn't invest any
work in changing the test environment to 2.1.5.

> > Also note that the latest SP release is 2.5.1 and that CentOS 4
> > generally is not a supported plattform for the SP.
> > Not that any of that will help with your problem,
> 
> Yes, I understand that I have to stay with the 2.4 release for our
> CentOS4 servers.

Not sure that aligns with the project's interpretation of "officially
supported software": Only RHEL/CentOS 5+6 are now supported
(independent of the SP release, AFAIU) and the 2.4 SP is no longer a
"previous stable" release per the SHIB2 home page in the wiki.

> Apart from yours and Scott's posts, I did also get a private email
> from a company offering me their paid IdP service creation, but
> nothing on whether my Jmeter tests might have a serious problem.

If that is a common pattern (i.e., if others have been approached in
similar ways) I'd request removing those from the list (if they even
are subscribed; the archives are public so that may not help much).
-peter