[asterisk-users] No MOH with parked call

[Steve Davies]

On 23 December 2010 18:01, Steve Davies <davies147 at gmail.com> wrote:
> Hi Again,
>
> I thought I had this sorted, but it appears that in a clean
> environment I did not in fact fix it. There appears to be a bit of a
> contradiction.
>
> 1) In 1.6.2.x, musiconhold requires DAHDI (which we have)
> 2) In 1.6.2.x parked calls get MOH only if res_timing_dahdi is not loaded...
>
> I am confused. MOH in general terms works just fine, but if I park a
> call with res_timing_dahdi loaded, I get silence after the orbit
> announcement. If I unload res_timing_dahdi, all works fine, but my
> timing sources are less reliable.
>
> I have backported res_musiconhold.c from 1.8 to 1.6.2.16-rc1, but this
> does not seem to fix things - is the problem elsewhere? Is there a fix
> that I can try, or perhaps backport?

Further to this, I have been slowly tracing through the codepath for a
parked call - ast_settimer is called correctly for the MOH generator,
and seems to set up the DAHDI timer exactly the same way as it does
for the alternative timing modules, and all of the setup calls return
success. I don't have a verbose output trace to hand, but the sequence
seems to be as follows:

-- Start moh_files_generator successfully using ast_settimer()
   <Chunks of MOH file are generated here>
-- Stop moh_files_generator successfully
-- Play the parking position numbers
-- Start moh_files_generator using ast_settimer()
<Generator is never called>

Any thoughts where to look next? It is odd that it appears to be
working immediately before the position is read out, but then fails to
restart afterwards...

Thanks,
Steve