See below (sent in July):

---
On a different note I believe to have located the source of CPU-hungry
boxes. The problem lies in Kannel's handling of HTTP 1.1 requests. Try using
the send-sms interface or the Kannel-admin module from a HTTP 1.1 compliant
browser, leave the browser window open and watch the bearerbox or smsbox go
powerhungry ~1 minute after the request. It happens every time!!

My temporary workaround has been to treat all HTTP-requests as HTTP 1.0 (in
http.c), but this is only a short-term solution as there are important
benefits from using HTTP 1.1.
---

This problem has existed for a long time and has never been fixed.
Unfortunately, while my workaround has severely helped, two simultaneous
HTTP-request to Kannel can sometimes have the same effect.

cheers,

Frederik Ammitzb�ll
Unwire
Vestergade 12A, 3.
1456 K�benhavn K

Tlf.:  +45 33 33 08 70
Mobil: +45 27 11 99 99
Fax :  +45 33 33 09 70
Web:   www.unwire.dk


> > I think this has to do with the way some of the SMSC's have been
> > implemented. The basic pattern is that the smsc code fires up a thread
> > or two, both of which rotate around an infinite loop. Depending on how
> > polling of incoming data is done, these loops may well consume a lot
> > processing power. This is what I noticed while implementing the smsc_cgw
> > code (or, I did precisely the same thing). I solved it by making the
> > threads sleep some small amount of time (with pollfd) while waiting for
> > data.
> >
>
> In my case this is smsbox eating cpu, not bearerbox.
>
>
>
> br
> R.
>
>
>
>


Reply via email to