Calls have started dropping recently

[b0bb1]

We've been using Asterisk for many years (12+ probably), and since 2012 we've been using Flowroute. Never had any issues until about a couple of months ago. Around the same time Flowroute discontinued their Nevada PoP and we reconfigured our Asterisk server to point to the US EAST VA PoP. I'm not sure if this is related to the issue though.

What's been happening recently is that calls just drop randomly, and this occurs more frequently with longer conference calls. Flowroute is telling us this:

"The issue was the rport was present initially in the invite and then on the 200 OK it was no longer present. Because of this, we (Flowroute) changed our signalling to the IP in the contact header. If the rport is kept throughout the call then we will continue to signal to the original IP and the call should not end. If you do change the configuration we can look at the call to see if that does fix the issue."

I'm not sure what configuration needs to change. We haven't touched the configurations for a very long time now (the only change was made to the host and fromdomain settings below (from sip.conf). As far as I know, the dropped calls issue had started occurring before we even changed these settings.

; ####################################################
; ## Flowroute Settings ##
; ####################################################

[flowroute]
type=friend
secret=xxxxxxxxxx
username=xxxxxxx
host=us-east-va.sip.flowroute.com
dtmfmode=rfc2833
context=inbound
canreinvite=no
allow=ulaw
allow=g729
insecure=port,invite
fromdomain=us-east-va.sip.flowroute.com


We also tried changing the nat=yes setting to nat=force_rport,comedia but that didn't help at all. The asterisk server is behind a firewall, and all the phones are in the same office behind the same firewall as well.

[general]
context=default
allowoverlap=no
srvlookup=no
jbenable = yes
jbmaxsize = 200
jbresyncthreshold = 1000
register => xxxxxxxx:[email protected]
externip=aa.bb.cc.dd
localnet=192.168.15.60/255.255.255.0
;nat=yes
nat=force_rport,comedia


Hoping someone can provide some assistance.

 

[dhoppy]

I think the advice you are going to hear in this forum is going to be "sorry, we don't support 12 year old systems, you should upgrade."

 

[billsimon]

"The issue was the rport was present initially in the invite and then on the 200 OK it was no longer present. Because of this, we (Flowroute) changed our signalling to the IP in the contact header. If the rport is kept throughout the call then we will continue to signal to the original IP and the call should not end. If you do change the configuration we can look at the call to see if that does fix the issue."

"rport" means they will signal back to you on the path you started with them. If you are behind a NAT then the source port coming from your firewall is probably randomized and they are responding to you on that port.

Not having "rport" mean they will use what is in your Contact header, which should be your external IP and standard SIP port (5060 probably unless you changed it).

In short, you need to adjust your firewall and SIP configuration to allow traffic back to you. If you search for flowroute threads you will see a lot of angst about their multiple IP addresses. It is time for you to upgrade to a newer system where you can implement the PJSIP stack and set up "match" rules for their range of IP addresses.

 

[b0bb1]

billsimon, thanks for your suggestion. I reviewed flowroute related posts and found this:

BCM One

support.flowroute.com

We had our firewall configured for /32, and changing it to /28 did help. Have not dropped any long calls since that change yesterday. I'm thinking that when there is a node change behind Flowroute's load balancer, it forces a reinvite and that was the cause of our dropped calls. But not sure, simply a guess at this point as we have not had dropped calls since we changed the whitelist to /28.

Hope this helps someone, and thanks for your reply!