[VoiceOps] DASH issues

Justin Shore justin at justinshore.com
Thu Nov 5 01:24:16 EST 2009

Is anyone else running into issues with DASH's SBC changes from this 
past morning's maintenance window?  We're experiencing two new issues.

The first is a DTMF issue.  We're seeing DASH send us RFC2833 DTMF over 
the SIP trunk but the MetaSwitch is not liking what the new DASH SBCs 
are sending.  In turn the MetaSwitch is not sending DTMF on to the 
downstream customer.  It sounds like RFC2833 negotiation between the 
MetaSwitch and their new SBC leaves only ulaw as an option which doesn't 
appear to agree with MetaSwitch.  Thus no DTMF on towards the customer. 
  We were moved back to the old SBC and the DTMF problem has been 
resolved for now.  We've been told that this was an issue earlier with 
DASH and MetaSwitch and that it was resolved at a particular MetaSwitch 
customer's site.  We're working back channels to try and figure out what 
that fix was.

The second issue is with what initially appears as a no-way audio on 
calls to certain numbers at different customer sites.  When you capture 
the traffic you realize that there is 2-way audio but it only contains 
comfort noise.  DIDs don't seem to be affected.  The only numbers chosen 
for this particular problem are the main numbers at the sites.  The 
sites in question all use the same remote media gateway model on the 
MetaSwitch, have Edgemarc CPEs and Aastra phones.  The lines are also 
configured for SIP forking with several phones on the same line. 
However it doesn't affect all of our customers configured in what appear 
to be the exact same way.  Some work perfectly fine.  What we're hearing 
from DASH is that their upstream is putting the call on-hold for some 
reason as soon as the receiving end picks up.  Very strange.  We're 
troubleshooting that one in the morning.

Has anyone else run into issues with the migration to the new SBCs at 
DASH?  Since we only had outgoing calls going over the new DASH SBCs and 
incoming on the old SBCs these problems weren't ever discovered in 
testing.  Without that there really wasn't any testing done on our end.


More information about the VoiceOps mailing list