The Micronets websocket proxy

The websocket proxy allows 2 websocket clients to connect to each other using outbound connections to a common URI.

1 Quick Start

1.1 Checkout out the Micronets websocket proxy project

From a directory containing your micronets repositories (e.g. "~/projects/micronets")

git clone git@github.com:cablelabs/micronets-ws-proxy.git

This will create a "micronets-ws-proxy" directory containing the websocket proxy.

1.2 Setting the proxy parameters

For now, the websocket proxy's parameters are stored in the bin/websocket-test-client.py source. If you're running the proxy on your local machine (for testing), you don't need to change the defaults - which should be something like this:

proxy_bind_address = "localhost"
proxy_port = 5050
proxy_service_prefix = "/micronets/v1/ws-proxy/"
proxy_cert_path = bin_path.parent.joinpath ('lib/micronets-ws-proxy.pkeycert.pem')
root_cert_path = bin_path.parent.joinpath ('lib/micronets-ws-root.cert.pem')

If you have a test server to run the proxy on, most likely the only param you would want to change is the proxy_bind_address - which you can set to 0.0.0.0 to listen on all interfaces, or set to the IP address of the interface you want to expose the proxy on.

1.3 Setting up the websocket proxy environment

Note that the Python websocket proxy requires python 3.6, pip, and the "virtualenv" package. All other dependancies are installed into the virtualenv for the websocket proxy.

The python virtualenv with the library dependancies is setup by performing the following steps:

From the micronets-ws-proxy directory:

virtualenv --clear -p $(which python3.6) $PWD/virtualenv
source virtualenv/bin/activate
pip install -r requirements.txt

1.4 Running the websocket proxy manually

From the python-infrastructure directory, run:

virtualenv/bin/python bin/websocket-proxy.py

You should see output similar to the following:

2018-12-10 12:56:07,310 micronets-ws-proxy: INFO Loading proxy certificate from lib/micronets-ws-proxy.pkeycert.pem
2018-12-10 12:56:07,312 micronets-ws-proxy: INFO Loading CA certificate from lib/micronets-ws-root.cert.pem
2018-12-10 12:56:07,313 micronets-ws-proxy: INFO Starting micronets websocket proxy on 0.0.0.0 port 5050...

The websocket proxy can be stopped via Control-C. But it will also be stopped if/when the terminal session the proxy is started from terminates. It can also be run via "nohup". But running it via systemd is the preferred method (see below).

1.5 Running the websocket proxy using systemd

An example systemd service control file micronets-ws-proxy.service is provided in the source distribution. The "WorkingDirectory" and "ExecStart" entries need to be modified to match the location of the websocket proxy virtualenv and python program. And the "User" and "Group" settings should be set to the micronets user (or commended out to run as "root") E.g.

WorkingDirectory=/home/micronets-dev/Projects/micronets/micronets-ws-proxy
ExecStart=/home/micronets-dev/Projects/micronets/micronets-ws-proxy/virtualenv/bin/python bin/websocket-proxy.py
User=micronets-dev
Group=micronets-dev

The systemctl service unit file can be installed for the systemd service using:

sudo systemctl enable $PWD/micronets-ws-proxy.service
sudo systemctl daemon-reload

Once the micronets-ws-proxy service is installed, it can be run using:

sudo systemctl start micronets-ws-proxy.service

Where the logging will be stored is system-dependent. On Ubuntu 16.04 systems logging will be written to /var/log/syslog.

The status of the proxy can be checked using:

sudo systemctl status micronets-ws-proxy

and the proxy stopped using:

sudo systemctl stop micronets-ws-proxy

1.6 Running the websocket proxy using Docker

1.6.1 To build the Docker image:

docker build --tag=micronets-ws-proxy .

1.6.2 To start a container locally and run it in the background:

On Linux:

docker run -d --network host --restart unless-stopped --name micronets-ws-proxy-service micronets-ws-proxy

On MacOS:

docker run -d -p 5050:5050 micronets-ws-proxy

1.6.3 Deploying a Docker image to Artifactory

A Makefile is provided to generate the Docker image and upload it to the configured artifact repository.

Both can be accomplished by running:

make docker-push

Note that the destination repository and path is configured in the Makefile and that Docker will request credentials in order to perform the push.

1.6.4 Retrieving the latest docker image from Artifactory

The commands to retrieve the latest Docker image(s) for the websocket proxy are also contained in the included Makefile.

To pull the latest Docker(s) run:

make docker-pull

Note that the source repository and path is configured in the Makefile. No credential should be required to pull the Docker image.

2 Setting up proxy authorization using certificates

This repository contains pre-generated certificates for testing/prototyping. For a discreet/production installation, new certificates should be generated and deployed. The instructions below detail a basic prodedure for generating new certificates.

2.1 Generating the shared root certificate used for websocket communication:

This will produce the root certificate and key for validating/generating leaf certificates used by peers of the websocket proxy:

bin/gen-root-cert --cert-basename lib/micronets-ws-root \
    --subject-org-name "Micronets Websocket Root Cert" \
    --expiration-in-days 3650

The shared root cert will be the basis for trust for all the entities communicating via the websocket proxy. The websocket proxy will only trust peers that present a cert (and can accept a challenge from) the proxy.

The micronets-ws-root.key.pem file generated by this script should only be retained for the purposes of generating new leaf certs for the websocket peers.

2.2 To generate the cert to be used for the Websocket Proxy:

bin/gen-leaf-cert --cert-basename lib/micronets-ws-proxy \
    --subject-org-name "Micronets Websocket Proxy Cert" \
    --expiration-in-days 3650 \
    --ca-certfile lib/micronets-ws-root.cert.pem \
    --ca-keyfile lib/micronets-ws-root.key.pem

cat lib/micronets-ws-proxy.cert.pem lib/micronets-ws-proxy.key.pem > lib/micronets-ws-proxy.pkeycert.pem

The lib/micronets-ws-proxy.pkeycert.pem file must be deployed with the Micronets websocket proxy and well-protected. The lib/micronets-ws-root.cert.pem must be added to the Proxy's list of trusted CAs. (and should really be the only CA enabled for the proxy)

2.3 Generating the cert to be used for the Micronets Manager:

bin/gen-leaf-cert --cert-basename lib/micronets-manager \
    --subject-org-name "Micronets Manager Websocket Client Cert" \
    --expiration-in-days 3650 \
    --ca-certfile lib/micronets-ws-root.cert.pem \
    --ca-keyfile lib/micronets-ws-root.key.pem

cat lib/micronets-manager.cert.pem lib/micronets-manager.key.pem > lib/micronets-manager.pkeycert.pem

The lib/micronets-manager.pkeycert.pem file must be deployed with the Micronets Manager to connect to the websocket proxy and lib/micronets-ws-root.cert.pem must be added to the Micronet's Manager CA list.

2.4 To generate the cert to be used for the Micronets Gateway Service:

bin/gen-leaf-cert --cert-basename lib/micronets-gateway-service \
    --subject-org-name "Micronets Gateway Service Websocket Client Cert" \
    --expiration-in-days 3650 \
    --ca-certfile lib/micronets-ws-root.cert.pem \
    --ca-keyfile lib/micronets-ws-root.key.pem

cat lib/micronets-dhcp-manager.cert.pem lib/micronets-dhcp-manager.key.pem > lib/micronets-dhcp-manager.pkeycert.pem

2.5 To generate the cert to be used by the test client:

bin/gen-leaf-cert --cert-basename lib/micronets-ws-test-client \
    --subject-org-name "Micronets Websocket Test Client Cert" \
    --expiration-in-days 3650 \
    --ca-certfile lib/micronets-ws-root.cert.pem \
    --ca-keyfile lib/micronets-ws-root.key.pem

cat lib/micronets-ws-test-client.cert.pem lib/micronets-ws-test-client.key.pem > lib/micronets-ws-test-client.pkeycert.pem

3 Testing the micronets websocket proxy

3.1 Downloading the micronets websocket proxy source (containing the test client)

mkdir -p ~/projects/micronets
cd ~/projects/micronets
git clone git@github.com:cablelabs/micronets-ws-proxy.git
cd micronets-ws-proxy
mkvirtualenv -r requirements.txt -a $PWD -p $(which python3) micronets-ws-proxy

3.2 Connecting the websocket test client to the proxy (using the same URI as a connected micronets gateway)

The websocket test client takes all its parameters via arguments. Use "-h" to see the options. A typical test session would look like this:

workon micronets-ws-proxy
bin/websocket-test-client.py --client-cert lib/micronets-manager.pkeycert.pem --ca-cert lib/micronets-ws-root.cert.pem  wss://ws-proxy-api.micronets.in:5050/micronets/v1/ws-proxy/test/gateway-0001

The test client should startup with log messages similar to:

Startup...
Loading test client certificate from lib/micronets-manager.pkeycert.pem
Loading CA certificate from lib/micronets-ws-root.cert.pem
ws-test-client: Opening websocket to wss://ws-proxy-api.micronets.in:5050/micronets/v1/ws-proxy/test/gateway-0001...
ws-test-client: Connected to wss://ws-proxy-api.micronets.in:5050/micronets/v1/ws-proxy/test/gateway-0001.
ws-test-client: Sending HELLO message...
ws-test-client: > sending hello message:  {"message": {"messageId": 0, "messageType": "CONN:HELLO", "requiresResponse": false, "peerClass": "micronets-ws-test-client", "peerId": "12345678"}}
ws-test-client: Waiting for HELLO message...

and when a peer is connected to the proxy:

ws-test-client: process_hello_messages: Received message: {'message': {'messageId': 0, 'messageType': 'CONN:HELLO', 'peerClass': 'micronets-ws-test-client', 'peerId': '12345678', 'requiresResponse': False}}
ws-test-client: process_hello_messages: Received HELLO message
ws-test-client: HELLO handshake complete.
MyHTTPServerThread.__init__(): state: ready
ws-test-client: Starting event loop...
ws-test-client: receive: starting...
MyHTTPServerThread: Starting HTTP server on localhost port 5001...

3.3 Using the websocket test client's built-in HTTP proxy to test the websocket

To use the websocket-test-client’s built-in server to test the proxying of an HTTP REST request and response to the Micronet’s DHCP server running on the gateway, requests can be sent to the port specified using the "--http-proxy-port" argument (or port "5001" if not specified).

e.g.

curl -H "Content-Type: application/json" http://localhost:5001/micronets/v1/dhcp/subnets

When sending a request via the built-in http test server, the websocket-test-client program will encapsulate the HTTP request sent by curl into a “REST:REQUEST” websocket message and send it to the proxy - which will send it to the gateway. After processing the request, the micronets-dhcp server will send a response, encapsulate it in a “REST:RESPONSE” websocket message, send it to the proxy - which will send it to the websocket-test-client program. The websocket-test-client program will then take the contents out of the websocket message and send the response to curl.

See below for details on the micronets websocket proxy message format.

4. API Keys

4.1 Generating the shared root certificate used for micronet API component communication:

This will produce the root certificate and key for validating/generating leaf certificates used by certain micronets API:

bin/gen-root-cert --cert-basename lib/micronets-api-root \
    --subject-org-name "Micronets API Root Cert" \
    --expiration-in-days 3650

This root cert is be the basis for trust for all the clients communicating via select Micronets APIs. APIs will only trust peers that present a cert signed by this root cert. The API server must have client cert verification enabled/required and lib/micronets-api-root.cert.pem must be added to the API endpoint's list of trusted CAs (and should really be the only CA enabled).

The micronets-ws-root.key.pem file generated by this script should only be retained for the purposes of generating new leaf certs for the websocket peers. It should not be deployed with any software components.

4.2 Generating API client certificates:

bin/gen-leaf-cert --cert-basename lib/micronets-api-client \
    --subject-org-name "Micronets API Client Cert" \
    --expiration-in-days 3650 \
    --ca-certfile lib/micronets-api-root.cert.pem \
    --ca-keyfile lib/micronets-api-root.key.pem

cat lib/micronets-api-client.cert.pem lib/micronets-api-client.key.pem > lib/micronets-api-client.pkeycert.pem

The lib/micronets-api-client.pkeycert.pem file must be deployed with any micronet software component that needs to access a cert-controlleed micronets API.

5. Websocket message format

5.1 Base message definition

All messages exchanged via the websocket channel must have these fields:

{
   “message”: {
      “messageId”: <client-supplied session-unique string>,
      “messageType”: <string identifying the message type>,
      “requiresResponse”: <boolean>
      “inResponseTo”: <id string of the originating message> (optional)
   }
}

5.2 HELLO message definition

{
    "message": {
        "messageId": 0,
        "messageType": "CONN:HELLO",
        "peerClass": <string identifying the type of peer connecting to the websocket>,
        "peerId": <string uniquely identifying the peer in the peer class>,
        "requiresResponse": false
    }
}

Example:

{
    "message": {
        "messageId": 0,
        "messageType": "CONN:HELLO",
        "requiresResponse": false,
        "peerClass": "micronets-ws-test-client",
        "peerId": "12345678"
    }
}

5.3 REST Request definition

This defines a REST Request message:

“message”: {
   “messageType”: “REST:REQUEST”,
   “requiresResponse”: true,
   “method”: <HEAD|GET|POST|PUT|DELETE|…>,
   “path”: <URI path>,
   “queryStrings”: [{“name”: <name string>, “value”: <val string>}, …],
   “headers”: [{“name”: <name string>, “value”: <val string>}, …],
   “dataFormat”: <mime data format for the messageBody>
   “messageBody”: <either a string encoded according to the mime type, base64 string if dataFormat is “application/octet-stream”, or JSON object if dataFormat is “application/json”>

Note that Content-Length, Content-Type, and Content-Encoding should not be communicated via the "headers" element as they are conveyed via the dataFormat and messageBody elements. If the request is handled by a HTTP processing system, these header elements may need to be derived from dataFormat and messageBody.

Example GET request:

{
  "message": {
    "messageId": 3,
    "messageType": "REST:REQUEST",
    "requiresResponse": true,
    "method": "GET",
    "path": "/micronets/v1/dhcp/subnets",
    "headers": [
      {
        "name": "Host",
        "value": "localhost:5001"
      },
      {
        "name": "User-Agent",
        "value": "curl/7.54.0"
      },
      {
        "name": "Accept",
        "value": "*/*"
      }
    ]
  }
}

Example POST request:

{
  "message": {
    "messageId": 1,
    "messageType": "REST:REQUEST",
    "requiresResponse": true,
    "method": "POST",
    "path": "/micronets/v1/dhcp/subnets",
    "headers": [
      {
        "name": "Host",
        "value": "localhost:5001"
      },
      {
        "name": "User-Agent",
        "value": "curl/7.54.0"
      },
      {
        "name": "Accept",
        "value": "*/*"
      }
    ],
    "dataFormat": "application/json",
    "messageBody": {
      "subnetId": "mocksubnet007",
      "ipv4Network": {
        "network": "192.168.1.0",
        "mask": "255.255.255.0",
        "gateway": "192.168.1.1"
      },
      "nameservers": [
        "1.2.3.4",
        "1.2.3.5"
      ]
    }
  }
}

Example PUT request:

{
    "message": {
        "messageId": 3,
        "messageType": "REST:REQUEST",
        "requiresResponse": true,
        "method": "PUT",
        "path": "/micronets/v1/dhcp/subnets/mocksubnet007",
        "dataFormat": "application/json",
        "headers": [
           {"name": "Host", "value": "localhost:5001"},
           {"name": "User-Agent", "value": "curl/7.54.0"},
           {"name": "Accept", "value": "*/*"}
        ],
        "messageBody": {
            "ipv4Network": {
                "gateway": "192.168.1.3"
            }
        }
    }
}

5.4 REST Response definition

This defines a REST Response message:

{
    “message”: {
        “messageType”: “REST:RESPONSE”,
        "inResponseTo": <integer message ID of the REST:REQUEST that generated the response>
        “requiresResponse”: false,
        “statusCode”: <HTTP integer status code>,
        “reasonPhrase”: <HTTP reason phrase string>,
        “headers”: [{“name”: <name string>, “value”: <val string>}, ],
        “dataFormat”: <mime data format for the messageBody>,
        “messageBody”: <either a string encoded according to the dataFormat, base64 string if dataFormat is         “application/octet-stream”, or JSON object if dataFormat is “application/json”>
 “application/octet-stream”, or JSON object if dataFormat is “application/json”>
    }
}

Example GET response:

{
    "message": { 
        "messageId": 2,
        "inResponseTo": 3,
        "messageType": "REST:RESPONSE",
        "reasonPhrase": null,
        "requiresResponse": false, 
        "statusCode": 200,
        "dataFormat": "application/json", 
        "messageBody": {
            "subnets": [
                {
                    "ipv4Network": {
                        "gateway": "192.168.30.2",
                        "mask": "255.255.255.0",
                        "network": "192.168.30.0"
                    }, 
                    "subnetId": "wireless-network-1"
                }, 
                {
                    "ipv4Network": {
                        "gateway": "192.168.40.1",
                        "mask": "255.255.255.0",
                        "network": "192.168.40.0"
                    }, 
                    "subnetId": "wired-network-3"
                },
                {
                    "ipv4Network": {
                        "gateway": "10.40.0.1",
                        "mask": "255.255.255.0",
                        "network": "10.40.0.0"
                    }, 
                    "nameservers": ["10.40.0.1"],
                    "subnetId": "testsubnet001"
                }
            ]
        }
    }
}

Example POST response:

{
    "message": {
        "messageId": 2,
        "inResponseTo": 1,
        "messageType": "REST:RESPONSE",
        "requiresResponse": false,
        "statusCode": 201
        "dataFormat": "application/json",
        "messageBody": {
            "subnet": {
                "subnetId": "mocksubnet007",
                "ipv4Network": {
                    "gateway": "192.168.1.1",
                    "mask": "255.255.255.0",
                    "network": "192.168.1.0"
                }, 
                "nameservers": ["1.2.3.4", "1.2.3.5"]
            }
        }
    }
}

Example PUT response:

 {
     "message": {
         "messageId": 2,
         "inResponseTo": 3,
         "messageType": "REST:RESPONSE",
         "requiresResponse": false,
         "statusCode": 200,
         "dataFormat": "application/json",
         "messageBody": {
             "subnet": {
                 "ipv4Network": {
                     "gateway": "192.168.1.3",
                     "mask": "255.255.255.0",
                     "network": "192.168.1.0"
                 },
                 "nameservers": ["1.2.3.4", "1.2.3.5"],
                 "subnetId": "mocksubnet007"
             }
         }
     }
 }

Example DELETE response:

{
    "message": {
        "messageId": 3,
        "inResponseTo": 5,
        "messageType": "REST:RESPONSE",
        "requiresResponse": false,
        "statusCode": 200
    }
}

5.5 Event Message definition

{
    “message”: {
        “messageType”: “EVENT:<client-supplied event name>”,
        “requiresResponse”: False,
        “dataFormat”: <mime data format for the messageBody>,
        “messageBody”: <either a string encoded according to the mime type, base64 string if dataFormat is “application/octet-stream”, or JSON object if dataFormat is “application/json”>
    }
}

jayd2446/micronets-ws-proxy