FXO not immediately disconnecting

[chris.berry.1]

Hello,

I have a client CUCME/CUE system that utilizes FXO ports for outside public lines. We are getting a fair number of "hang up" voicemails (3 seconds into the voicemail you can hear the disconnect) as the FXO ports seem to take a few extra seconds to disconnect after the caller has hung up. I'm looking for advice or solutions on fixing this.

Kind regards,

Chris                  

[Harmit Singh]

Hi Chris,

Classic case of supervisory disconnect issue. Did you get a chance to review this link yet:

http://www.cisco.com/en/US/tech/tk652/tk653/technologies_tech_note09186a00800ae2d1.shtml

HTH.

Regards,

Harmit.

[chris.berry.1]

Hi Harmit,

I did take a look at that. My issue most closely matches scenario 1 in the linked article. Am I understanding correctly that I should try implementation of Supervisory Tone Disconnection with timers set at 5 for both wait-release and call-disconnect?

FXO_Paper#

configure terminal

FXO_Paper(config)#

voice-port 3/1/1

FXO_Paper(config-voiceport)#

supervisory disconnect dualtone mid-call

FXO_Paper(config-voiceport)#

cptone us

FXO_Paper(config-voiceport)#

timeouts wait-release 5

FXO_Paper(config-voiceport)#

timeouts call-disconnect 5

FXO_Paper(config-voiceport)#

exit

Thanks,

Chris

[Harmit Singh]

Hi Chris,

Yes, that is worth a try, it should ideally work. Let me know how it goes. Secondly, the default cptone is always US. Is the router located in the US or elsewhere? The frequency and cadence values for supervisory disconnect vary depending on the cptone specified.

We can always create a custom cptone and specify the frequency and cadence for disconnect tone as given by your service provider. However, usually they aren't very cooperative in that respect :-) Which means a PCM capture would be needed. I can then do a stream analysis after decoding the dat file and suggest the values for the custom cptone.

Here is a previous thread on such an issue:

https://supportforums.cisco.com/message/3069056

HTH.

Regards,

Harmit.

[chris.berry.1]

Hi Harmit,

I was just able to test this and it did not help. The call takes 11 seconds from external caller hang-up until the FXO port goes on hook.

Thanks,

Chris

[paolo bevilacqua]

Actuall 11 seconds is better than never.

Post relevant config.

[chris.berry.1]

voice-port 0/1/0

trunk-group ALL_FXO 64

translation-profile incoming INCOMING_CallerID_PROFILE

supervisory disconnect dualtone mid-call

timeouts call-disconnect 5

timeouts wait-release 5

connection plar opx 198

caller-id enable

IOS 15.1(4)M4b

[paolo bevilacqua]

Try reducing call-disconnect and wait-release to 1 or 2.

[Harmit Singh]

Hi Chris,

Did you do a shut/no shut on the voice-port after adding those commands? If not, please do so. Else, like Paolo rightfully suggested, try reducing it down further. If all else fails, then we look at making a custom cptone. If telco can help give you the disconnect frequency and cadence values then great, else we'll need a PCM capture (instructions in the post I had shared with you earlier).

HTH.

Regards,

Harmit.

[chris.berry.1]

Hi All,

I have reduced call-disconnect and wait-release to 2, as well as shut/no shut - the problem persists. I will attempt to engage the telco and provide an update after.

Thanks,

Chris

[chris.berry.1]

The telco (AT&T) claims they do not have this information (frequency and cadence). I want to mention these are centrex lines using loop start. They told me we could change to ground start (for a fee of course). Should we proceed with the PCM capture?

Thanks,

Chris

[paolo bevilacqua]

The US tones are well know. Beside the tones are right, otherwise the port woul not disconnect at all.

You can stay with loop-start and timers reduced to the minimum, or upgrade to ground-start or digital circuits. The latter are the preferred solution, together with SIP.

[chris.berry.1]

I believe the timers are already very minimal and there was no change - disconnects still took 11 seconds even after the change. Ground start would eliminate this issue with certainty?

The customer will question why the "old" phone system worked without ground start - any insight in that regard?

[paolo bevilacqua]

Correct tone-based disconnect works in seconds, not minutes, before you have mentioned 11 seconds, and that is kind of acceptable.

I don't know why it doesn't work in you case, you can begin listening if a disconnection tone is actually sent at all.

All circuit types except loop-start FXO provide reliable disconnect supervision.

[chris.berry.1]

I incorrectly stated minutes - I meant seconds. It takes 11 seconds for the disconnect.