SPA122 SIP484 response causes excessive NOTIFY event: reserve-trunk

[George Spanos]

I've noticed when the SPA122 dials a call and receives a SIP 484 "Address Incomplete" response from the call agent, the call agent plays an announcement "call cannot be completed as dialed" and then the SPA plays what sounds like a dial-tone. It's actually playing the "Prompt tone" instead of a re-order or busy signal type of tone. During the "prompt tone" the SPA is sending out extremely excessive "NOTIFY event: reserve-trunk" messages to the call agent. Within 5 seconds it had send 25x NOTIFY messages and received 25x 404 not found messages. I have never seen the event "reserve-trunk" nor has google....

I have observed this issue under the newest firmware version 1.3.1-003 and the old 1.2.1-004 release. It sure sounds like this mass amount of NOTIFY messages sent out could've added to the catatonic-ness of v1.2.1-004.

Here's the part of the call flow where I observed the behavior:

21:46:19.871105 - From Metaswitch Call Agent              

<---------------------------------------------

SIP/2.0 484 Address Incomplete

Call-ID: xxxx-xxxxxxx10.11.15.24

CSeq: 102 INVITE

From: "xxxxxxxxxx" <sip:xxxxxxxxxx@callagent.

voice.com>;tag=xxxxxxxxxxxxxx

To: <sip:xxxxxxxxxx@callagent.voice.com>

;tag=sip+xxxxxxxxxxxxxxx

Via: SIP/2.0/UDP 10.11.15.24:5060;

received=10.11.15.24;branch=xxxxxxxxxxxx

Content-Length: 0

Supported: resource-priority

Contact: <sip:10.1.81.180:5060>

Server: DC-SIP/2.0

Organization: MetaSwitch

21:46:19.925989 - From SPA122

--------------------------------------------->

ACK sip:xxxxxxxxxx@callagent.

voice.com:5060 SIP/2.0

Via: SIP/2.0/UDP 10.11.15.24:5060;branch=xxxxx

From: "xxxxxxxxxx" <sip:xxxxxxxxxx@callagent.

voice.com>;tag=xxxxxxxxxx

To: <sip:xxxxxxxxxx@callagent.

voice.com>;tag=sip+xxxxxxxxxxxxx

Call-ID: xxxxxxxxxxxxx@10.11.15.24

CSeq: 102 ACK

Max-Forwards: 70

Authorization: Digest username="xxxxxxxxxx",re

alm="callagent.voice.com",nonce="xxxxxxxxxxxx

",uri="sip:xxxxxxxxxx@callagent.voice.com

:5060",algorithm=MD5,response="xxxxxxxxxxxx

",qop=auth,nc=xxxxxx,cnonce="xxxxxxx"

Contact: "xxxxxxxxxx" <sip:xxxxxxxxxx@10.11.15.24

:5060;ref=xxxxxxxxxx>

User-Agent: Cisco/SPA122-1.3.1(003)

Content-Length: 0

21:46:20.225619 - From SPA122

--------------------------------------------->

NOTIFY sip:callagent.voice.com:5060

SIP/2.0

Via: SIP/2.0/UDP 10.11.15.24:5060;branch=xxxxxx

From: "xxxxxxxxxx" <sip:xxxxxxxxxx@callagent.

voice.com>;tag=xxxxxxxxxxxx

To: <sip:callagent.voice.com>

Call-ID: xxxxxxxxxx@10.11.15.24

CSeq: 20699 NOTIFY

Max-Forwards: 70

Event: reserve-trunk;call-id=xxxxxxxxxxxxxxxxx

@10.11.15.24;expires=10

User-Agent: Cisco/SPA122-1.3.1(003)

Content-Length: 0

21:46:20.227915 - From Metaswitch Call Agent 

<---------------------------------------------

SIP/2.0 404 Not Found

Call-ID: xxxxxxxxxxxxxx@10.11.15.24

CSeq: 20699 NOTIFY

From: "xxxxxxxxxx" <sip:xxxxxxxxxx@callagent.

voice.com>;tag=xxxxxxxxxxxx

To: <sip:callagent.voice.com>;tag=sip

+1+xxxxxxxxxxxxxx

Via: SIP/2.0/UDP 10.11.15.24:5060;received=10.11.

15.24;branch=xxxxxxxxxxxxxxxxx

Content-Length: 0

Server: DC-SIP/2.0

Organization: MetaSwitch

Has anyone else noticed this? I see this as an issue because it plays the wrong tone and it sends a mass amount of messages to the Call Agent that don't really apply to anything. Why are tnese being sent at all? What does "reserve-trunk" really do? Maybe it was meant for UC540 or CME??

I have a TAC case open for this issue.

Message was edited by: George Spanos

[Dan Lukes]

Just for completeness - I tried to replay such scenario and I discovered that loop doesn't occur when the NOTIFY is rejected by "489 Bad event" response from PBX (it's more appropriate response for unknown events, of course, device should not flood network in any case). And/or "expires=0" in original NOTIFICATION may affect it.

Dan

<<<

SIP/2.0 484 Address incomplete

From: "6018" ;tag=8773d4554ea7044fo0

To: <11>;tag=as13c4f5f8

Call-ID: a775a14d-91eadbb7@A.B.12.246

CSeq: 102 INVITE

Server: Asterisk PBX 1.8.20.1

Allow: INVITE, ACK, CANCEL, OPTIONS, BYE, REFER, SUBSCRIBE, NOTIFY, INFO, PUBLISH

Supported: replaces, timer

Reason: Q.850;cause=28

...

>>>

NOTIFY sip:A.B.250.1:5060 SIP/2.0

From: "6018" ;tag=d0e0a73c447c60bco0

To:

Call-ID: cdbf2592-efcf1c85@A.B.12.246

CSeq: 8816 NOTIFY

Max-Forwards: 70

Event: reserve-trunk;call-id=a775a14d-91eadbb7@A.B.12.246;expires=0

User-Agent: Cisco/SPA112-1.3.1(003)

Content-Length: 0

...

<<<

SIP/2.0 489 Bad event

From: "6018" ;tag=d0e0a73c447c60bco0

To: ;tag=as1965bbda

Call-ID: cdbf2592-efcf1c85@A.B.12.246

CSeq: 8816 NOTIFY

Server: Asterisk PBX 1.8.20.1

Allow: INVITE, ACK, CANCEL, OPTIONS, BYE, REFER, SUBSCRIBE, NOTIFY, INFO, PUBLISH

Supported: replaces, timer

Content-Length: 0

...

[George Spanos]

Dan, that's interesting. I agree the device shouldn't go nuts and flood the network of course. Our Metaswitch call agent sends that SIP 404 and in our logs it shows something like "Username not found." I can't find anything on google or Metaswitch support docs about that type of NOTIFY event.

If I could find some details maybe I could tell the call agent to reply with the 489 as you've shown.

[Dan Lukes]

I can't find anything ... about that type of NOTIFY event. 

Like me. I assume it's a Cisco proprietary extension. I suspect it has something to do with SIP to PSTN gateway and overlap dialling handling.

[Patrick Born]

Hi George,

I've checked with my colleagues in Interop, QA, and Engineering. Long-story-short, we have interop tested with the 8.1 Metaswitch stream and believe this issue to be related to an interaction issue between the ATA and a patch in the 7.4 Metaswitch stream.

We have developed a test release [1.3.2(004)] of firmware which I can share with you if you'd like. I'm at paborn@cisco.com if you're interested in trying it.

Regards,

Patrick--

[Patrick Born]

I should have added that this issue is assigned to CDETS# CSCue45076

Regards,

Patrick---

[George Spanos]

Just to keep this thread up to date I'm adding my email reply to TAC & Patrick here also.

We’re using Metaswitch v7.4. I just tested the new release fw v1.3.2-004 and I see a few issues with it.

1.) Line 1 doesn’t work at all, I see no registration attempts and the light is completely off. Line 2 does work/register, light is on, I’m able to make calls.

2.) I setup the same scenario as before and I’m still hearing the dial-tone-like sound and seeing tons of NOTIFY messages being sent.

I don’t see any fixes in this version, actually breaks line 1.