On Thu, 2009-11-26 at 13:13 +0000, Gabor Paller wrote:
> At that point the load test tool was already idle - it did not generate
> any calls because it was already stopped. The last call was generated at
> 11:36, that particular entry was at 11:39, 3 minutes later.
> 
> Immediately after the last call, the transaction count was higher, 1962
> (similar log entry). Then it started to decrease gradually as more and
> more transactions were closed.

There are a number of things to be remembered in this situation.  One is
that there are a lot of SipTransaction objects for each call.  The other
is that SipTransaction objects are removed from the system via a
timeout, IIRC, 3 minutes.  So if you are doing load testing, you expect
to see a *lot* of SipTransaction objects, many times the number of calls
that are active at a given moment.  You also expect that it will take a
while for the SipTransaction objects to be garbage-collected out of the
system.  If the only component in the system is the proxy, you expect
the number of SipTransaction objects to go to zero, but that may take 5
or 10 minutes.  If that doesn't happen, there may be a problem with
garbage-collection.

In regard to testing your load generator, this is not a particularly
good test.  No matter how deranged your load generator is, the number of
SipTransaction objects at a given moment should be limited, and if the
load stops, that number should go to zero over a few minutes.  A better
test is to verify that each and every call that was originated reached
its destination, that all the proper responses were generated, etc.

Dale


_______________________________________________
sipx-users mailing list [email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-users
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-users
sipXecs IP PBX -- http://www.sipfoundry.org/

Reply via email to