Alison Reidy schrieb:
> I’m trying to find out if we can throttle the messages per second rate
> that Kannel submit’s to the provider.   
> 
> Also, related to this is that our SMSC disconnects us when we’ve
> exceeded our throttle, and Kannel stores these messages that it can’t
> send.  At this point it seems, , kannel’s Virtual memory just increases
> until the OS causes the system to crash.
> 
>  So the problem is two fold: when kannel can’t send messages (because
> the SMSC’s throttle has been exceeded), it’s holding them and trying
> again, and that retry seems to have a memory leak which causes the
> system to crash.

there is no memory leak in Kannel as far as we know currently. We use valgrind
and other (own internal debug malloc wrappers) to test this.

If you see an increase in the VIRT memory segment for the daemons, then a) your
queues increase drastically over time and b) Linux allocates considerable large
chunks for the stack space of each thread.

So if you use a lot of SMSC connections you will see typically 10M per thread
just for the thread start. Each thread has like a normal process, a stack space
and a heap. The stack space is STATICALLY pre-allocated, the heap is DYNAMICAL.

So, no mem-leaking, unless you can prove it with some valgrind output ;)

Stipe

-- 
-------------------------------------------------------------------
Kölner Landstrasse 419
40589 Düsseldorf, NRW, Germany

tolj.org system architecture      Kannel Software Foundation (KSF)
http://www.tolj.org/              http://www.kannel.org/

mailto:st_{at}_tolj.org           mailto:stolj_{at}_kannel.org
-------------------------------------------------------------------

Reply via email to