Charmed Quark Systems, Ltd. - Support Forums and Community

Full Version: New Amazon Echo Support
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
A global action, setting the GVar:EchoReply variable, but it doesn't get that far.

The text coming in doesn't match the filters I have setup in the Echo file (guest mode on, guest mode off, etc) with "mode on". I am pretty sure its on the Amazon end not CQC.
Amazon has apparently commissioned some college students to conduct a survey of reactions to Echo. There are a opportunities to state things you would like improved.

They state the results will be read by Senior Echo Tech people. I picked this up over on the "Un-official Amazon Echo Forum.

Whenever possible while filling it out, I suggested improved interfacing with 3rd party home automation software.

http://www.echotalk.org/index.php/topic,229.0.html
I did it, and provided a fairly detailed explanation of where it falls short for our needs. The biggest, and probably least intrusive, change they could make would be to allow you to just configure the Echo to always use a specific skill, and avoid having to use the extra verbiage. That couldn't be THAT difficult to do.

Other improvements I mentioned probably aren't likely, given that they would require a fairly fundamental change. Ultimately those would be better served I guess by supporting something like the Kinect.
Trying to get this setup and failing when alexa tells cqc to load the configuration.

For my setup, I have a single Server2k12R2 machine setup to handle the SSL termination but it also does reverse proxying. So, for example, a request to https://myhostname.com/ goes to my web server but a request to https://myhostname.com/cqcecho goes to my cqc server (hitting the cqc server just at its root or whatever follows cqcecho).

I setup the URL in the javascript to be myhostname.com/cqcecho. Is that valid? Or is there something else I need to do to support a path in there?
There may be an issue with the encrypted connection. Give it a try unencrypted first, just as a sanity check. If that works, then you know it's something to do with encryption, else you know it's something more run of the mill.

Where are ayou adding the /cqcecho? Are you putting that on the host name at the top? Depending on how the Javascript is building up the URL, that might not work, I dunno. Or if it's using that value somewhere down there as a literal host name, maybe that's an issue.

Wouldn't you be better off maybe just using a different port and setting up separate port forwards?
I added a reverse proxy (F5 LTM) as well, that eliminated whatever the SSL negotiation issues are with CQC.

Did you change the calls to HTTPs in the Javascript file (not just at the top) but down where the acutal calls are made? You can't just change the URL you need to use an HTTPS object to make the HTTPS calls (i.e. https.xxx vs http.xxx)
The latest Javascript looks at the HTTP/HTTPS setting at the top and calls either require('http') or require('https') accordingly, so it shouldn't require any modifications now down in the code.

When you say you added a reverse proxy, what does that mean? How is this dealing with the issue? That might indicate what the issue actually is.
A reverse proxy terminates the client TCP connection and creates a new one to the server. In-between it allows you to manipulate whatever you want all the way up to the application layer.

I am terminating the SSL conversation on the front end then re-sending the traffic unencrypted to the server (SSL Offload). Unfortunately it doesn't really help determine what the actual SSL negotiation issue is since I am not re-encrypting to CQC.

The path looks correct, matches what I have with the exception of case "/CQCEcho". I have seen the CQC web server be sensitive to case in URLs in the past so you might check that too.
I dunno what's going on. If you can make an HTTP browser connection to the CQC WS, then it's got to be dealing with the encrypted session correctly, and AFAIK everyone can make such a connection.

Is the proxy inside your network? If so, at least you are covered externally, so that's good.

Does it have any sort of log of what is transpiring between it and the AWS server? That could be potentially useful.
Yeah it's secure, it's inside. It's actually more secure since it will filter things and do some vulnerability and DOS protection, before anything touches the server.

There isn't really anything informative in the logs. Unfortunately SSL is a pretty complex beast it has a couple of layers of negotiations to get the tunnel up. And these days security is forcing both sides to be stricter on what they accept.

It you are just instantiating an MS SSL object of some sort then it may be an issue in my windows 7 configuration that windows doesn't like.

I am still learning all the in's and out's of SSL type issues and I haven't seen the issue before nor did Google turn up much. If I remember right CQC was resetting on its' side due to something Amazon was wanting to do.