wscat | curl for websockets | Websocket library

Issue was with data returned from handlers which has to be in string format using JSON.Stringify and output returned from handler has to be in specific format.

You have most likely forgotten to deploy your API (Routes -> Actions -> Deploy API). WebSocket APIs still use the old console which does not have auto deploy.

It works when I try it

Websockets are disabled by default. See RSKj configuration reference

To enable websockets on RSKj: (1) Add -Drpc.providers.web.ws.enabled=true to your java command when starting RSKj. (2) Update the configuration file. The file is named mainnet.conf for RSK Mainnet.

• See RSKj node configuration for how to locate and edit the config file.
• See RPC protocol configuration reference for a detailed explanation about the various RPC configurations available, including the ones related to websockets.

Your config file should look like this:

Found the issue. My serverless.yml had an extra space of indentation for the disconnect and message lambda definitions. No error. Deployed no problem...but never worked. I guess I'll pay attention to it next time...spent like 3 hours on it. Facepalm!

It seems like you're setting 1 lambda to handle 2 trigger sources, one is IoT service, the other is API Gateway Websocket. Since you use 1 lambda, you have to handle cases when the request is came from sources:

1. While event.requestContext is available when the request is triggered from API Gateway, it is not available when the request is triggered from IoT service (check the IoT event object here https://docs.aws.amazon.com/lambda/latest/dg/services-iotevents.html). So the error you faced (which is Cannot read property 'domainName' of undefined") is about that. You should turn off the lambda trigger from IoT service or handle the request when it comes from IoT Service.
2. I'm not sure about the forbidden error but it is more like you sent unstructured message to API gateway WS, it should be connection.send(JSON.stringify({ action: "sendmessage", data: "hello world" })); instead of connection.send("hello world");

Edited based on post update:

I know ws is there because if I console it it returns a big object with a bunch of functions

Lambda function is not really a server, it is an instance Node environment (that's why it is called FUNCTION), Lambda function doesn't work as the way you think normal Nodejs app does, its container (node environment) usually is halted (or freeze) whenever its job is done so you cannot keep its container alive like a normal server. That's the reason while you can console log the Websocket object, you cannot keep it alive, the NodeJS container was already halted whenever you return/response.

Since you cannot use the Websocket object to open WS connection in Lambda, Amazon offers a way to do that via API Gateway. The way we work with API Gateway Websocket is different than the normal server does too, it would be something like:

• User -> request to API Gateway to connect to websocket -> call Lambda 1 (onconnect function)
• User -> request to API Gateway to send message over Websocket -> call Lambda 2 (sendmessage function)
• User -> request to API Gateway to close connection -> call Lambda 3 (ondisconnect function)

3 settings above is configured in API Gateway (https://docs.aws.amazon.com/apigateway/latest/developerguide/apigateway-websocket-api-integrations.html), logic of 3 functions onconnect, sendmessage, ondisconnect can be handled in 1 lambda or 3 lambda functions depending on the way we design, I check your 3 lambda functions and it looks okay.

I see that you want to use IoT but I'm not sure why. You should test your Websocket API first without anything related to IoT. It would be better if you can tell what you want to achieve here since IoT works more like a publish/subscribe/messaging channel and I don't think it's necessary to use it here.

I got an aswer from the ws github issues, so I am posting it for anyone else

Turns out this line

endpoint := fmt.Sprintf("https://%s/%s/@connections/%s", domain, stage, connectionID)

needs to turn into this

endpoint := fmt.Sprintf("https://%s/%s/", domain, stage)

A common trick to use process substitution to only wait for the last stage in the pipeline:

Ambassador will redirect incoming request froms http to https.

You can verify by