Hi Magnus,
I think you can simply the problem by consider this. If your clients
"randomly" select a web server to talk to and they do this using a flat
distribution, you could assume that you will get a 1/3 split of clients
going to each webserver. The easy way to manage this is to divide your
clients in 3 groups and have each group hit a different server.
Regards,
Kirk
Magnus Olstad Hansen wrote:
Hello,
I'm trying to stress test a server net where we have 3 webserver
nodes. Loadbalancing is done by our client side application (that is,
each client randomly selects which of the webnodes it'll talk to). To
test this correctly I wanted to create the same behavior in JMeter -
but preferrably without setting up my rather long list of request once
per each server. So I thought the following test plan would do the
trick (see the lines with *):
- Test Plan
-- Thread Group
--- Random Controller *
---- HTTP Request Defaults (www1) *
---- HTTP Request Defaults (www2) *
---- HTTP Request Defaults (www3) *
--- HTTP Get
--- HTTP Get
--- HTTP Post
--- Loop
(blah blah blah)
However - this does nothing. Not a single request is sent. I did not
understand much until I suspected that no server were set by the "HTTP
Request Defaults" config elements in the random controller. So I tried
inserting another "HTTP Request Defaults" before the Random Controller
- and it works.
I suppose the problem is that the config block only applies to its
parent level (ie the Random Controller). Any tips on how to get a
setup like this to work?
Thanks for your time!
Magnus
---------------------------------------------------------------------
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]