If it returns back to normal, then it's technically not a leak :-).
However, the event queue based design of the sipX C++ libraries and
applications could indeed result to an ever increasing queue size if it
is given continuous load so i am not surprised if memory grows in a
predictable linear pattern if you give it a load test. We have an
internal soak test framework managed by Douglas and 4.4 is the first
version that tasted its wrath. I am not sure if Douglas has the
statistics handy. Lets see if he chimes in.
On 01/05/2012 05:11 PM, Henry Dogger wrote:
Hi all,
When testing a setup for one of our clients we came across a
disturbing issue.
We generated a high load of about 180 calls per minute on the system
via ISDN, and wanted to let this run for let's say a week.
But after one night we noticed the sipXproxy is not returning its
memory or sipXproxy is not doing this fast enough...
When we shut the load test down, memory usage was returned back to
normal in about an hour.
Is this normal behavior? A server under this kind of load will
eventually start using its swap memory and then failing (we tested this).
At first we tested with a light machine (only 2Gbyte of memory), but
we get the same results with a machine with a bigger CPU and 6 Gbyte
of memory, but the time it takes the system to fail is longer...
Can anyone shed some light on this?
We are running: sipXecs (4.4.0- 2011-10-12EDT17:34:54
domU-12-31-39-00-0D-21)
Kind regards,
Henry Dogger
Telecats BV
_______________________________________________
sipx-users mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-users/
_______________________________________________
sipx-users mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-users/
