More on Troubleshooting SpeakerTrack 60

This is not a failure prone device, its far from it but when you do have an issue because it does not happen very often it can be difficult to find troubleshooting information. Below are some tips and tricks I have seen and used to help diagnose and fix issues.

Issue: Audio tracking and face detection not aligning

This is a much harder condition to diagnose because it looks like its working but seems really slow to change camera angles and zoom in on the face detected.

The best way to discover this issues is with these two commands:

xCommand Experimental SpeakerTrack Diagnostics Start

xCommand Experimental SpeakerTrack Diagnostics Stop

Once your in diagnostic mode what you will notice is a lot of red squares which means a face has been detected as shown below but not a lot of green which means a voice has not been matched to a face. I also noticed that F value was pegging at around 100%.

clip_image004[4]

Solution: Firstly you should always open a TAC Case but in saying that you can also factory reset the SpeakerTrack unit (not the codec). This can help to recover the SpeakerTrack and restore functionality.

Refer to page 17 of the SpeakerTrack Admin guide how to do this:

http://www.cisco.com/c/dam/en/us/td/docs/telepresence/endpoint/camera/precisionhd/user_guide/precision60-speakertrack60-camera-guide.pdf

For the MX700 Dual the picture below points to where the reset button is for the camera modules. Handy to know if you have a single camera that goes offline. The ST unit also has a similar factory reset button. This can also be done via a null modem cable connected to the serial port but most people do not often have a modem cable handy and a paper clip is much easier to find.

SpeakerTrack FR

More Handy Commands for SpeakerTrack:

Checking status:
xstatus // speakerTrack
*s Cameras SpeakerTrack Availability: Available
*s Cameras SpeakerTrack LeftCamera VideoInputConnector: 1
*s Cameras SpeakerTrack RightCamera VideoInputConnector: 2
*s Cameras SpeakerTrack Status: Active
xstatus // Peripherals
*s Peripherals ConnectedDevice 1045 Name: "SpeakerTrack 60"
*s Peripherals ConnectedDevice 1045 SoftwareInfo: "MT7.3.0.8cb420c"
*s Peripherals ConnectedDevice 1045 Status: Connected
*s Peripherals ConnectedDevice 1045 Type: SpeakerTracker
Things to look for are whether ST is connected and also the software version number aligns with your codec versioning.
Turning on/off SpeakerTrack:
xCommand Cameras SpeakerTrack Activate
xCommand Cameras SpeakerTrack Deactivate
SpeakerTrack Setup document:

http://www.cisco.com/c/en/us/support/docs/collaboration-endpoints/telepresence-speakertrack-60/118733-config-speaker-00.html


Previous Troubleshooting post:


http://voipnorm.blogspot.com/2015/08/troubleshooting-cisco-telepresence.html


If you have run into other commands that are handy or tips and trick's please feel free to post in the comments.


VoIPNorm

Troubleshooting Cisco TelePresence SpeakerTrack 60

More and more companies I am working with are deploying Cisco TelePresence SpeakerTrack 60 with either a MX integrated device or as a standalone with an SX80. So I thought it would be good to let everyone in on a very helpful command to watch the SpeakerTrack in action. There are a couple of steps involved to turn on the diagnostic mode and it can only be done via the command line today.

If you have never used SSH to log in to a Cisco TelePresence device your going to have to download and install a SSH client like Putty or from a MAC use the command line. Here is the link to Putty: http://www.putty.org/

SSH into the codec via its IP address.

image

Log in.

Run the following commands to turn SpeakerTrack diagnostic mode on and off.

xCommand Experimental SpeakerTrack Diagnostics Start

xCommand Experimental SpeakerTrack Diagnostics Stop

Below is a screenshot of diagnostic mode turned on and a face is detected but no audio match.

clip_image004

Now the engineer is speaking below, and the green square indicates a positive match for facial detection and voice. Therefore the system will zoom in to this person.

The indicators down the bottom of the screen match up with the following:

F = 10.1% detected voice
T = 91.6% non-noise
E = 0% voice from far end
C = 28.1% camera movement
U = 0% ultrasound detected
N = 89.9% silence
S = 178 samples from sound algorithm

clip_image006

If you run this it will help you understand if SpeakerTrack is working but you will need someone in the room to see it displayed on the screen and hae a face to actually track. You wont see jack if you do not have someone in the room with the SpeakerTrack!

Here is the API reference guide just in case your curious on other commands:

http://www.cisco.com/c/dam/en/us/td/docs/telepresence/endpoint/sx-series/tc7/api-reference-guide/sx80-api-reference-guide-tc71.pdf

VoIPNorm

Jabber and CUCM TFTP Server Site Selection

CUCM TFTP server selection for Cisco phones has been a well known architecture for quite some time. The phone looks up a TFTP server address based on DHCP option 150 and downloads its configuration and files from the published address. Simple enough. Jabber on the other hand works a little differently as it does not use DHCP directly. The underlying PC takes care of addressing and Jabber has to find alternative methods to get to its TFTP server to download its configuration file.

In most cases Jabber uses a SRV records to locate a CUCM server running the User Data Service (UDS). UDS is enabled by default on all your CUCM servers by the way. Once this is located a service profile is returned which has a number of things, but in our case we are most interested in is the TFTP server addresses. Depending on how many TFTP servers you have enabled in your environment your service profile file will list all of them automatically. So if you have turned on TFTP on ever server in your cluster that’s what it will return.  See slide below for a more graphical look at the process.

 

image

Initially this may seem like a good idea to turn TFTP on every CUCM server and you have redundancy but this isn't how Jabber reads the file. Jabber will check to see if the first three TFTP servers are available only and if all three of those TFTP servers happen to be in a data center that is off line the client will not be able to pull down the jabber-config.xml file.

So what is the best practice? Never have more than two CUCM TFTP servers per cluster enabled per data center for Jabber. This will ensure that Jabber will never be stuck looking for a third server in the same datacenter and will need to look else where, hopefully in another datacenter where you have a third TFTP running. This doesn’t affect how a phone retrieves its information though because a phone as we mentioned earlier is using DHCP option 150. This also doesn’t affect how many TFTP servers you can have in your cluster just where you turn the service on if your also running Jabber and what high availability. In saying that ideally three TFTP servers is probably the optimal amount you will want or need there is nothing stopping you having more. I would suggest that if your desire is more than three have a solid engineering reason behind it.

As always, I hope this helps some hapless sole looking for an answer.

VoIPNorm