[asterisk-dev] Problem with Asterisk realtime ?

Oded Arbel oded at atelis.net
Mon Oct 23 10:14:08 MST 2006


On Fri, 2006-10-13 at 16:54 +0200, Oded Arbel wrote:
> On Thu, 2006-10-12 at 14:15 -0300, Joshua Colp wrote:
> > Oded Arbel wrote:
> > > Funny thing though - I was sure that it was working correctly this
> > > morning.. I did restart the asterisk several times, just to see if this
> > > would cause it to behave, but no luck. I just tested and other settings
> > > (such as 'secret' and 'name') don't refresh either. Adding a new sip
> > > peer to the realtime database does work immediately without a reload.
> 
> > Do you have realtime caching enabled? If so the database won't be 
> > touched again until it disappears from memory.
> 
> Yes, I have. I understand that voicemail MWI won't work otherwise ? Is
> it possible to set caching to be less aggressive - maybe clean entries
> from the cache after a timeout ?

I've disabled 'rtcachefriends' in sip.conf, and all was good for a
while. Then I wanted to add DUNDi into the mix - I'm using 'regcontext'
to have users added automatically to some dialplan context, and then
have DUNDi lookup on that. 

Unfortunately, when 'rtcachefriends' is disabled, Asterisk still says
that new SIP users are added to the context:
<log>
-- SIP Seeding peer from astdb: '333' at 333 at 62.90.116.111:5066 for 3600
-- Added extension '333' priority 1 to dundi-priv-local
</log>
but they are not really added, or at least 'dialplan show' doesn't show
them in the specified context. DUNDi apparently doesn't think such
automatically registered extensions exist either, as it fails to look
them up. Enabling 'rtcachefriends' again fixed that issue, but now I'm
back to the first problem above - changes to database records
necessitate a full 'sip reload'. 

I'm still thinking one of these behaviors is a bug - what do you think?


--
Oded Arbel
Atelis
oded at atelis.net
Tel: +972-54-7340014
::..
"If I had my life to live over, I'd try to make more mistakes next
time."
    -- Jorge Luis Borges




More information about the asterisk-dev mailing list