We have a rare intermittent issue where a customer is speaking with a CCX agent and that customer also hears a completely different conversation occuring between another CCX agent/customer. The calls are SIP. I've searched around the forums, the bug lookup tool, etc. and have basically found that IOS/DSP bugs can potentially cause this to happen. As well, I have seen someone suggest it is a provider issue. I am guessing since there aren't many confirmed resolutions on the support forums that it's difficult to track down.
We're running IOS 15.0(1)M4 on a 2951 router. DSP version is showing 26.3.8. I did identify a couple of crosstalk-like behaviors with this version in the bug lookup tool (CSCtk34885 and CSCty33945), but unfortunately the behavior is not identical. Although, just knowing that similar behaviors do in fact exist and are addressed with repaired IOS images is enouraging.
Does anyone have any suggestions or wisdom to share?
One of the things you can check is the IPCC Extension of the agents. Make sure that on CUCM > Call Routing > Route Plan Report if you search the IPCC Extension, you get only 1 result. This is because the IPCC Extension has to be unique. Also on the IPCC Extension line configuration, make sure that it has maximum number of calls set to 2 and maximum busy trigger to 1.
On the UCCX side, make sure that the script has 2 second delay at the beginning and before the accept step. This is just a recommendation in order to make sure that the system will be able to connect the channels while getting a new call.
If the configuration is okay, this will require UCCX logs, cucm traces and possibly a triggered PCM capture from the gateway for an RCA.
Thanks. This particular agent does have 4 max calls and busy trigger of 2. I am curious though, how does that cause the issue? What about the IOS/DSP being a possible cause?
This is a standard practice and is even documented in Cisco troubleshooting documents like the one below
Regarding IOS / DSP being the possible cause i haven't come across any known bugs based on the information available. However, i would suggest trying changing the max calls and busy trigger settings first and if the issue persists then logs / traces / captures need to be collected as per my initial response.
I understand it is standard practice, but it seems completely unrelated. There is no documentation I can find that claims the behavior could be related to max lines/busy trigger. Oh well, it's a mystery that nobody can answer. Thanks anyway.