Siptrunk problem UPDATE - different scenario's

[Fr0gg3r - MaartenS]

Hi All,

A bit a last resort question here as Avaya does not want to help without a DECT support contract and the provider is pointing at the pbx as the issue.

We have a customer with multiple sites, same setup everywhere.
IP Office R11 + DECT installation + siptrunk of Proximus (Belgium)

Only on 1 site we have a problem with DECT phones calling to mobile phones.

50% of the calls are OK​

25% of callsetups are timedout after 30 seconds.​

²25% of calls are OK but only after 30 seconds.​

IP phones,digital phones do not have the problem.

We placed a test pbx +dect installation at site A where the problem occurs. Result = Also our test pbx has this problem​

We placed the same test pbx+dect installation at site B. Result= the problem does not occur there.​

-> this rules out that the pbx/dect is the problem because same pbx works on site B but no on site A.
-> looking at this test it would suggest that this is a provider problem at site A.

But how do we explain why this issue is not happening when using an ip or digital phone on this location.
-> this suggests that the dect installation is the problem.

Avaya wants a dect support contract to investigate, Provider says problem is the pbx

Analysis: all calls were picked up immediately by the mobile phone so no delayed media should be present.All calls were made from same user to the same mobile phone
When a DECT phone calls out to a mobile phone we have 3 possible scenarios:
[ol 1]
[li]Scenario1: Call is good[/li]
[li]Scenario2: Call is bad:

dect phone dials out -> mobile phone immediately takes the call. DECT phone keeps hearing ringback tone for +30seconds even though the mobile phone has already taken the call. The mobile phone does not hear anything. After 30 seconds the media path is established and we have 2way audio.​

[/li]
[li]Scenario3:Call is bad:

same scenario but after 30 seconds we receive a 408 request timeout from the provider and the call is dropped.​

[/li]
[/ol]

Below are the call scenarios from left to right (scenario 1-2-3)
S1: good call: pbx receives 183, pbx send media after it.
S2: bad call: pbx receives 183,pbx does not send media after the 183. pbx receives an update from provider but pbx does not respond to it. After 30 seconds the pbx receives an 200OK answer on the initial invite and the media is being send in both directions
S3: bad call: pbx receives 183,pbx does not send media after the 183. pbx receives an update from provider but pbx does not respond to it. After 30 seconds the pbx receives 408 request terminated from the provider.

So 3 question immerge here: bearing in mind that these scenarios do not occur on the same test pbx at a different location:
Why does the pbx not respond by sending RTP stream on the 183 session in progress?

In call 1 the pbx sends RTP​

In call 2 and 3 it does not. Same destination/source as call 1​

subquestion: Is it mandatory that the pbx responds to the 183 by sending rtp like in scenario1​

Why does the pbx not respond on the UPDATE SDP messages in scenario 2 and 3?

subquestion: Must the pbx respond to the UPDATE message from the provider before the provider forwards the 200OK respons from the initial invite?​

​

The provider told us they hold back the 200OK respons on the initial invite because the pbx did not respond to the UPDATE message. The update message does not contain any additional info compared to the 183 that was already received. The update is actually a converted 'ringing' which the provider converts for some reason.​

Why does the provider respond differently on scenario 2 and 3 while the flow is exactly the same.

We are very convinced that our setup is 100% correct. We have compared our config to the IP Office installations that the provider itself used to install. Siptrunk config is the same.
We have the same installations running on other locations and customer where no problems occur.

Thank you anyone for helping us out.

 

[G van Hamburg]

Can it be that good calls have the IP address in the from field and bad calls a domain? Do you have an SBCE in between? Or a router with SIP ALG active? To me, I’d say the update message is ignored by the PBX and or it never comes to the PBX.

Freelance Certified Avaya Aura Engineer

 

[Johnhyde]

What version of IP Office and Dect software are you using?

 

[Fr0gg3r - MaartenS]

@G van Hamburg:
- from headers are all the same in good or bad calls
- No SBCE between IPO and provider SBC. The provider delivers the siptrunk in LAN, it does not run via public internet.
- No router between ipo and provider that we control.

- The update message is in fact received by the ipo. We can see it in the monitor but the IPO does nothing with it.
We also see the retransmissions of the UPDATE from the provider in the sysmon.
After each retransmission that is received we see in the monitor:

13:25:39 6845576mS Sip: SIPDialog TXN : Decoding of message Succeeded 0 (retransmission)​

@JohnHyde:
We are at R11.1.0.2
DECT = IPBS2[10.4.6 C], Bootcode[10.4.6], Hardware[IPBS2-Z3A/1A]

 

[G van Hamburg]

Ok, and does the 200OK reach the IPO too? And in IPO, can you disable prack support?

Freelance Certified Avaya Aura Engineer

 

[Fr0gg3r - MaartenS]

If you are asking about the The 200OK response on the initial invite:

call 1: yes​

call 2: yes but only after 30 seconds because the provider is waiting for our pbx to first answer the UPDATE they send before they reply on the invite​

call3: no after 30 seconds the provider drops the call, they did not send a 200ok​

all 3 calls were from the same dect to the same number.

I see the update messages being received at the pbx.
Only in call1 the pbx acts on this.
In call 2 and 3 the pbx does nothing with the received update



We can disable PRACK as a test. I'll put it on the todo when we are back on site.

 

[G van Hamburg]

Can you send me a direct message? You can find me on linkedin.

Freelance Certified Avaya Aura Engineer