CUCM 6.1 and T.38 through SIP Trunk Provider

[bastienbertout]

Hi,

My customer is migrating is ToIP infrastructure from traditionnal gateway with E1s to a SIP Trunk.

I have to implement the design below :

VG224 <---sccp---> CUCM 6.1.4 <---sip trunk----> Provider

and

SagemXmediusFaxServer <---H323---> CUCM 6.1.4 <---sip trunk----> Provider

The Provider supports T.38 but it does not work with my VG224 or with the sagem fax server.

With my old E1 design, everything was working fine :

VG224 to/from 2821+E1 OK in T.38

Sagem to/from 2821+E1 OK in T.38

When I try to trace an incoming fax to VG224 or Sagem, I see that the provider sends me a SIP EO SDP with T38 capabilities but the CUCM answers with a 488 error : Media not accepted.

CUCM seems to be able to allow T.38 between H323 and MGCP devices but not through SIP Trunk.

Is there somewhere a matrix that could help me to validate what should work or not ?

Is there a way to bypass this problem (MTP, Cube or something else) ?

Thanx for your help,

Bastien.

[Chris Deren · Accepted solution]

Vg224 does support protocol-based T.38 which is what you need with H323/MGCP/SIP but not SCCP (SCCP only supports NSE based T38). So, you need to convert your Vg224 devices to use one of these protocols and ensure you enable T.38 on all GWs via:

fax protocol t38 version [0|3] ls-redundancy 0 hs-redundancy 0 fallback cisco

** version 0 configures legacy T.38 G3 fax support, whereas version 3 enables the newly supported SG3 over T.38 feature [as of 15.1(1)T]

HTH,

Chris

[bastienbertout]

oups, bad click about the rating ...

And bad info in my post, my VG224 is not in SCCP but in MGCP

Actually, protocol based T38 is already activated on my VG224. Here is en show mgcp

MGCP T.38 Max Fax Rate is 14400 bps

MGCP T.38 Fax is ENABLED

MGCP T.38 Fax ECM is ENABLED

MGCP T.38 Fax NSF Override is DISABLED

MGCP T.38 Fax Low Speed Redundancy: 0

MGCP T.38 Fax High Speed Redundancy: 0

So I'm looking for a T38 compatibility issue on CUCM between MGCP VG224 and SIP Trunks but I can't find any compatibility matrix on CCO.

Regards,

Bastien.

[Ayodeji Okanlawon]

Hi can you send a sh run of your voice gateway...

Can you also confirm that this is what your scenario looks like..

vg224--------mgcp--->CUCM----sip---Voicegatewy(CUBE)-----sip------SIP Provide?

can you also send the ff: debug?

Debug ccsip message

debug voip vtsp all

debug fax relay t30 all-level-1

[bastienbertout]

Hi,

Thanx for your answer.

There is no CUBE. CUCM is directly connected to provider's SBC through a SIP Trunk.

Regards,

Bastien.

[Ayodeji Okanlawon]

That means the only way you can troubleshoot this is to look at cucm traces...

Can you attach detailed CUCM traces? Can you also send a sh run of the vg224?

[bastienbertout]

For Now, I'm not on customer's site so I can't attach CUCM traces.

But, I can give you

- show run of the VG224 (see ConfVG224CG60.txt)

- A wireshark trace taken from sagem server (H323) for an incoming fax call through GW2821+E1 (H323) which works well : See

FaxViaT2.pcap

- A wireshark trace taken from sagem server (H323) for an incoming fax call through SIP SBC which does not work : See

FaxViaSBC-SFR.pcap

- A wireshark trace taken between the CUCM and the SIP SBC  for an incoming fax call through SIP SBC which does not work : See

SIPTrace.PCAP (you will see the error 488 not acceptable media)

109.0.3.20 is the provider SIP SBC

172.30.10.111 and 11.111 are Sagem Fax servers

172.30.10.22 is CUCM

172.20.33.2 is a 2821 GW

172.30.20.23 is a 2821 GW (Why do I see a RTP packet to this GW when it should be direct to the SBC ? FYI, this GW is not configured as a MTP in CUCM. It is configured as an IOS CFB)

Thanx for your help,

Regards,

Bastien.

[Chris Deren]

Glancing at your VG224 configuration you do not have protocol based T38 properly configured, here is the required configuration:

mgcp package-capability fxr-package

mgcp default-package fxr-package

Note: These two commands are needed to go from the default configuration of NSE-based T.38 to protocol-based T.38. The command no mgcp fax t38 inhibit is required for both NSE and protocol-based T.38, but it is the default.

HTH,

Chris

[bastienbertout]

Thanx for your help.

I made the changes but the behavior is still the same.

[Ayodeji Okanlawon]

Bastien,

Your mgcp configuration needs to be adjusted for NSE based T.38.

You also have modem passthrough configured...So please amend your configuration as follows..

no mgcp modem passthrough voip mode nse

mgcp package-capability pre-package

mgcp sdp simple

mgcp fax t38 ecm

mgcp fax t38 nsf 000000

mgcp fax t38 gateway force

You also need to disable fxr-package otherwise cucm will try to use protocol based T.38 and sicne Vg224 (with sccp) only support NSE based T.38 you want to disable all fxr-package ..

no mgcp default-package fxr-package

no mgcp package-capability fxr-package

[bastienbertout]

Thanx.

I already tried this but the SIP provider does not support NSE based T.38....

[Ayodeji Okanlawon]

Ok,

so yoru provide ronly support Standard Based T.38. Lets recap..and revert back to standard based t38. This is a working config for that can you compare your config and see what is miising..

Here is your present config and highlighted in red should be removed and blue should be added

mgcp

mgcp call-agent 172.30.10.22 2427 service-type mgcp version 0.1

mgcp dtmf-relay voip codec all mode out-of-band

mgcp rtp unreachable timeout 1000 action notify

no mgcp modem passthrough voip mode nse

mgcp package-capability rtp-package

mgcp package-capability sst-package

no mgcp package-capability res-package

no mgcp timer receive-rtcp

mgcp sdp simple

mgcp fax rate 14400

no mgcp rtp payload-type g726r16 static

mgcp default-package fxr-package

mgcp fax t38 ecm

mgcp fax t38 nsf 000000

[bastienbertout]

Hi,

I made the changes and still have the same behavior.

488 error : Media not accepted in SIP traces sent by CUCM to SIP Provider's SBC when this one announce T.38 capabilities.

best regards,

Bastien.

[Ayodeji Okanlawon]

Bastien,

It is quite difficult to troubleshoot a problem like this without looking at CUBE traces or CUCM traces. Since CUBE is not in your control, we can look at CUCM traces. This is the only way we can troubleshoot this. SO if you can enable detailed sip traces and send them over, it will help

[bastienbertout]

Thanx for your help.

Finally, I found that, for an unknown reason, all calls were redirected to a transcoder by CUCM.

So, I configured an MRG/MRGL with no transcoder and associated it to the trunk.

Now, all calls are OK between endpoint and SIP provider's SBC (without MTP or TC).

And everything works fine (G711 and T38).

One thing very important to know (I didn't and I've lost a lot of time) is that a TC or MTP ressource not associated to a MRG is available to all endpoints !

I had one TC ressource not associated to a MRG and the SIP trunk was using it even though the MRG/MRGL associated to the trunk did not have a TC ressource.

So don't forget to associate unused ressources to an "unused MRG" !!

Best regards,

Bastien BERTOUT.