Delay on speech path on SIP

[jamie77]

Calls come in on dedicated DSL from ITSP to IP500 R4.2. Most of the time it's all good. They take a few hundred calls a day on the SIP trunks.

Intermittently, the call will be presented to the phones, and when they answer they get dead air. If they hang on to the call, the speech path will eventually kick in and then it is all good. most of the time the caller will hang up and when/if the call back, the call will be fine.

The DSL also runs a VPN to a remote site with 56xx phones and IP DECT which are always fine. DSL circuits at both ends are provider (Voiceflex UK) managed with Cisco routers.

I have been working with the provider getting things changed at thier end to improve things, but it is slow going.

anyone come across this??

plea don't ask about upgrading from R4.2. That is a huge can of worms that would swallow us all!!!!!!!!!!!

Jamie Green

Football is not a matter of life and death-It is far more important!!!!

 

[Bas1234]

Could it be that someone does a FTP transfer from a remote site over the VPN which could hangup the router?

___________________________________________
It works! Now if only I could remember what I did...

Dain Bramaged
___________________________________________

 

[Duffman88]

Do you have a SIP or RTP/RTSP ALG enabled on the firewall?

Is this something that you can repeat or catch so you can get a packet trace inside & outside the firewall? I would like to see if the ITSP is sending its RTP stream on time and the firewall is causing the problem or if it's the ITSP not sending the media right away. Of course it could be something else, but in my experience I would look at those two traces first.

Chris
ACA- Implement IP Office

 

[Duffman88]

Is it a dedicated DSL circuit just for the SIP trunks?

Chris
ACA- Implement IP Office

 

[jamie77]

FTP transfer??? Now I'm wondering about the CCC wallboard on the other end of the link. Don't think it's FTP but it is the only non voice product on it.

It is a dedicated DSL from the SIP ITSP. It's very random and may not happen every day. I have the provider tracing it.

Jamie Green

Football is not a matter of life and death-It is far more important!!!!

 

[hairlessupportmonkey]

or could the issue be further up the chain like in BT's IP Exchange?

ask the ITSP to run some wireshark traces! if they wont, ditch them as they arent worth jack and email me ;-)

ACSS - SME

 

[Pepp77]

Jamie

Are you on sip.voiceflex.com?

If so ask them to move the customer to sip17.voiceflex.com

We had exactly the same issue with a customer (on version 5) with voiceflex and it was sip.voiceflex.com causing it, moving to sip17.voiceflex.com (or even sip11.voiceflex.com) resolved the issue.

We have now been about 2 months without the issue since the change.

| ACSS SME |

 

[Duffman88]

Hopefully Pepp77's suggestion fixes it for you otherwise those packet captures will let you know if it's your problem or your providers and give you proof. I would suggest getting your own packet traces as well as the provider getting them. Might be my craziness, but I just don't trust the providers or anyone for that matter when they say they are doing something. I like to see for myself.

Chris
ACA- Implement IP Office

 

[Duffman88]

Also if QoS is properly configured on the routers other traffic should not be interfering with the SIP trunks. I would also make sure your switches have QoS properly configured. I don't think this is the problem, but might as well cover all bases.

Chris
ACA- Implement IP Office

 

[jamie77]

I have no access to the routers. They are manged by Voiceflex.

Pepp, we they were on sip.voiceflex.com and moved about a month ago to sip17.voiceflex.com. It seemed to cure the issue but has now come back.!!!

VF have given me a graph apparently showing the improvement in service after changes to the DSL SNR made. just shows ping rates averaging about 35ms, which is fine. I have asked them to show how this proves the routing and RTP getting to site as it should.

I need to find a spare lappy and leave it on site with wireshark i think.

Jamie Green

Football is not a matter of life and death-It is far more important!!!!

 

[Duffman88]

I would definitely set up wireshark. I would also make sure IPO monitor is running with sip traces on as well so you can reference SIP, SDP and RTP information.

Chris
ACA- Implement IP Office

 

[jamie77]

Just need to find time inbetween the four install I have in the next week!!!

Jamie Green

Football is not a matter of life and death-It is far more important!!!!

 

[Duffman88]

Yea that's usually how it works and as soon as you start the packet traces the phones will be fine. This way you get to dig through tons of trace to find the problem call.

Makes you wanna smack the next person that goes "they are just phones why don't they just work?" haha.

Chris
ACA- Implement IP Office

 

[IPOfficeIreland]

I had it some time back on a site with 4.1/4.2 call centre. it was in fact 2 things that caused it. dtmf mode being passed through the trunk (call centre was a provider themselves so no rtp, qos issues)was not set to rfc2833 and the codec was g711u law. the dtmf issue caused 2 vcm channels to be used per call and the g711u law caused intermittant no speech issues. The delay before speech was the ipo negioating rtp codecs. locking down both in and out trunk to g711a and matching on ipo fixed that part. might not be the same in your case but worth checking.

 

[hairlessupportmonkey]

Jamie, seriously, get voiceflex to do the digging!

ACSS - SME

 

[edmana]

I have a customer using 1608 phones and is complaining that when they call other people's extensions internally and that person is on the phone, it just rings and eventually goes to VM.

They indicated their "old" system (not Avaya) gave them an indication the person was on a call.

Anyone know if this can be set up on a IPO 500 v1 running 6.0.18?

Thanks!
Ed

 

[Duffman88]

Edmana start a new thread for new questions.

Turn call waiting on for the user to allow another call in. I cant remember if it notifies the calling party if they are on the phone. I'm pretty sure it will show on the display. You can also use "ringback when free" to have the system set up a call between the two parties once they are available.

Chris
ACA- Implement IP Office

 

[Duffman88]

Hairlesssupportmonkey, first awesome name haha. It sounds like he has already been getting a bit of the runaround on the issue. So unless the ITSP is forced into seeing definitive proof (packet traces) the problem is on their end they will most likely keep sending their awesome ping statistics (hahaha) and keep blaming his network or equipment. It sucks but it sounds like Jamie is in the finger pointing game and the providers train just for occasions like this.

Chris
ACA- Implement IP Office

 

[hairlessupportmonkey]

Id say it would be easier for voiceflex to perform decent wireshark traces rather than the poor engineer having to set up a mirrored port on a switch, or install a hub and sit there grabbing packet traces.

A decent ITSP would already have this in place!

ACSS - SME

 

[Duffman88]

I fully agree. I constantly battled ITSPs and CLECs on things like this. No one ever wants to admit its their problem and tend to look at the surface instead of digging into it to find a solution. Luckily I found an ITSP that works really well with me and it's amazing how fast trouble resolution is now.

Chris
ACA- Implement IP Office