Documentum with Loadrunner

29 views
Skip to first unread message

Raul K

unread,
Jun 27, 2011, 3:37:21 PM6/27/11
to LoadRunner
Hello,
I am working with Documentum for a while now. I was successful to
automate most of transaction with vugen, check-in, check-out, import,
export. No issues with UID. I obviously used the documentum kit
provided by EMC to do it. In my company we use XMetal(Clinet based,
works with http protocol) and webtop. What all is possible by Webtop
can be done by XMetall too, all you have to do it choose Win 32
application instead of web based while recording it.

My problems are:

1.The kit provided is tested only for the http protocol and not https
protocol. I couldnt even record the script with https. So recorded the
script with http protocol, made it work and then replaced http with
https. Inorder to replay this script you have to enable Winlnet
instead of win soc.

2. I couldnt understand the part were the item is selected to check-in/
check-out. I got the content items IDs and used the search
functionality to get the item and then performed the checkin/checkout.
The problem here is when I run this for multiple iterations in vugen I
dont see any problem, but when I run it in the controller the script
fails. It searches the item, but the checkin/checkout fails, it is not
a UID problem because the script fails before it reaches there. Also
when ever this script starts in the controller, the application stops
responding and my load test fails. The CPU of the app server/ content
server are below 70%. When I perform the load test without these
scripts they look good.

I am an experienced user the Loadrunner and I wouldnt go wrong with
the run time settings, parameter option(unique, sequential etc..)

First of all I am sorry for a long mail. If you actually read the
whole thing, please let me know if you had come across anything like
this. My main concern here is, in most of the Web based application we
unselect the Winlnet option. If we choose this option then we have to
run the test as a process and not as a thread. But I run it as a
thread as I have only one sytem were the UCF setup is allowed. Could
this be a problem. But I think its because of the search functionality
I use in my script to do the checkin checkout.


Please get back to me if you can help me in anyways, also reach out to
me with your concerns also. I have been working on this for awhile now
and should be able to share my experience.

Thanks,

Raul

James Pulley

unread,
Jun 27, 2011, 3:57:18 PM6/27/11
to lr-loa...@googlegroups.com
1) There is nothing in HTTPs which mandates the use of WinInet. What
mandates this is when Windows Integrated security is in play. You would
then have a corresponding set of statements to set user identity to go along
with WinInet. I could see how if you replay under load without a
corresponding setting of identity for integrated security how this might
violate some business rules for number of concurrent sessions per user.
You might well have the situation that the ID assigned to the load generator
agent also doesn't have great credentials on the system under test, leading
to another source of conflict. As per the conversation on HTTP vs HTTPs
this morning, you may want to be cautious about the wholesale replacement of
one with the other as most optimal sites have a balance of these calls and
to SSL everything places an undue burden on the processor on the Web Server
without a whole lot of benefit beyond the encryption of PII/Financial data.

2) Generally whenever I see this situation I go back to the foundations.
Record twice and diff the recordings. Odds are there is something else
that is dynamic that requires some correlation for correct execution. As
to the early failure, checking for the content of returned pages is
recommended so you know where the business process comes off of the tracks.
It could be something as simple as test data in the incorrect state for the
intended use or something session related from multiple users of the same
name logging in, with the second instance getting a different view than the
first. Once you know where precisely it is failing the why should fall out
fairly naturally from what is observed.

My problems are:

Thanks,

Raul

--
You received this message because you are subscribed to the Google
"LoadRunner" group.
To post to this group, send email to LR-Loa...@googlegroups.com
To unsubscribe from this group, send email to
LR-LoadRunne...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/LR-LoadRunner?hl=en

Raul K

unread,
Jun 27, 2011, 7:03:30 PM6/27/11
to LoadRunner
Thanks James for the very quick response. I am involving another
person(Lr expert) to see what I have done. Hopefully another pair of
eyes can spot what I have missed. Will look into the user identity
that you mentioned. Will post soon with my findings.

Thanks

On Jun 27, 3:57 pm, "James Pulley" <loadrunner-li...@jamespulley.com>
wrote:
> For more options, visit this group athttp://groups.google.com/group/LR-LoadRunner?hl=en- Hide quoted text -
>
> - Show quoted text -

Ganesh Moorthy

unread,
Jun 27, 2011, 10:38:44 PM6/27/11
to lr-loa...@googlegroups.com
I am still at loss for the fact that it runs on Vugen on multiple
Iterations, but not in controller...
Are you sure the RequestID and SerialID are all captured and passed
properly for the frames?
During runtime what is the error you are getting in the controller (snapshot)?

another way is to monitor the WDK logs by only running the problem
script in controller and debug...

Reply all
Reply to author
Forward
0 new messages