[asterisk-dev] chan_h323 status in 1.8 branch ?

[Pavel Troller]

> In article <20110302120436.GA27212 at tangens.sinus.cz>,
> Pavel Troller <patrol at sinus.cz> wrote:
> > Hi!
> >   I was playing a little with chan_h323 in 1.8 SVN branch (version:
> > Asterisk SVN-branch-1.8-r309084M). It seems it has a lot of issues.
> 
> Judging by the activity in svn.commits, the chan_ooh323 module in addons
> is being much more actively maintained. It also has the advantage of
> being self-contained, rather than dependent on huge external C++ libraries
> of particular versions.
> 
> You might well have more success trying chan_ooh323 instead.
> 
Hi Tony,
  many thanks for your advice. I will definitely test chan_ooh323 too, but the
version in 1.6 addons is much less stable than 1.6 version of chan_h323.
I'm running H.323/SS7 gateways with a lot of traffic (about 20 CAPS and 500 
concurrent calls in the busy hour) and chan_ooh323 always crashed in less than
a minute. chan_h323 also crashes, but MTBC (mean time between crashes :-) ) is
about one week, which is much better. chan_ooh323 suffered (in the 1.6 version)
from a bad recovery from malformed/buggy received IEs - it obviously ended with
corrupted memory, overwriting some vital variables and then crashing soon. 
chan_h323 never caused a crash itself, it was always in h323plus/ptlib code
(most frequently ASN.1 parser - for the same reasons as chan_ooh323).

With regards, Pavel