[Tsung] Configuration Problem

[Matthias Stiller]

Thank you.

Just to guarantee that I understood everything right please consider the
following (senseless) szenario. I am using Tsung 1.2.1 ...

Let's say I want to produce 10 users (xmpp) in one second - and just these
10 users (maxusers = 10 ). :)
Every session lasts about 10 minutes with the following operations:
1. connection, 2. auth, 3. initial presence 4. roster:get (as described in
rfc 3921),  .... and now many presence and message stanzas ... and finally
closing the connection.

As far as I can see just the the connection establishment, the
authentication and the roster:get operations are producing response times.
The message and presence stanzas are just to produce a specific load, right?
After the test I will probably get a graph ("Requests and connection
establishment") with a (well, very small) peak at the very beginning and a
sloping curve to response times of 0 mseconds. 0 msecdonds because the
presence and message stanzas at the end are not considered as being
requests, right? So the green requests curve in the "Requests and connection
establishment"-graph just contains packets with ack-type local? So,
collecting response times during the whole tests is just possible when I
have some operations in my test szenario where I get a response from the
server (synchronously). For example roster:add. So having a roster:add or
something like this in the session might be very useful for analysing the
response time in general ... ?!

Thanks in advance and sorry for asking so many questions. :/



2008/4/24 Nicolas Niclausse <nicolas at niclux.org>:

> Matthias Stiller ecrivait le 17.04.2008 15:53:
>
>
>  I wonder how the response time in Tsung for every request (including
>> presence and messages) is being measured. Since XMPP is not synchronous you
>> don't get any answer to your messages. Is the receiving entity "waiting" for
>> an incoming message to calculate the response time? Or how is this done? As
>> far as I understand your acknowledgement system ack=local doesn't make sense
>> for presence and message stanzas ... ?! So the ack-attribute is not the
>> answer to my question, or?
>>
>
> Oops, I forgot to answer to this part of your message.
>
> response time is not measured for 'no_ack' mesage (this is true since the
> 1.2.2 version) because it doesn't make sense indeed (the rate of outgoing
> 'no_ack' requests is available though, see request_noack in the doc.).
>
> And yes, ack=local does not make sense for presence or message .
>
>
>
> --
> Nicolas
> _______________________________________________
> Tsung-users mailing list
> Tsung-users at lists.process-one.net
> https://lists.process-one.net/mailman/listinfo/tsung-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.process-one.net/pipermail/tsung-users/attachments/20080424/b1582578/attachment.html