hasura-connect

Hasura connector with MQTT hasura-connect cli helps us to connect MQTT broker with hasura graphql engine securely just with 2 commands.

This project also supports eclipse-sparkplug payload specification. set MODE='spBv1.0' to enable this.

• hasura-connect
• Usage
• Commands
• Configuring Hasura GraphQL Engine.
• Roadmap

Usage

$ npm install -g hasura-connect
$ hasura-connect COMMAND
running command...
$ hasura-connect (-v|--version|version)
hasura-connect/0.3.3 darwin-x64 node-v14.2.0
$ hasura-connect --help [COMMAND]
USAGE
  $ hasura-connect COMMAND
...

Demo

Commands

• hasura-connect connect
• hasura-connect help [COMMAND]
• hasura-connect init

hasura-connect connect

Connect to the MQTT broker

USAGE
  $ hasura-connect connect

OPTIONS
  -d, --debug  pass true to enable debugging

DESCRIPTION
  ...
  This will start an MQTT client and start subscribing to the topic specified in the config.json. every message payload 
  that is recieved will be parsed using getMutation method.
  getMutation method will be invoked with message and topic and the graphQL query generated by the method used to hit 
  hasura instance.

  "hasura-connect connect" will use the configuration from config.json by default, also it use getMutation method from 
  the parse.js to convert the payloads into necessary mutations. 
  considering the stateless behaviour of the runtime this can be executed parallelly in multiple instances, you may also 
  use process managers like pm2 to manage multiple process in the same machine.

See code: src/commands/connect.js

hasura-connect help [COMMAND]

display help for hasura-connect

USAGE
  $ hasura-connect help [COMMAND]

ARGUMENTS
  COMMAND  command to show help for

OPTIONS
  --all  see all commands in CLI

See code: @oclif/plugin-help

hasura-connect init

init command used to intialise hasura connect configuration file at the current directory

USAGE
  $ hasura-connect init

OPTIONS
  -c, --MQTT_CHANNEL=MQTT_CHANNEL  MQTT channel
  -h, --HASURA_HOST=HASURA_HOST    hasura host url
  -m, --MODE=MODE                  mode : sparkplug or normal
  -m, --MQTT_HOST=MQTT_HOST        mqtt host url

DESCRIPTION
  ...
  init will create a config.json and a parser.js on the current working directory, this allows the developer to specify 
  the configuration and to parser.js can be used as a custom parser which will return mutation & variables which will be 
  used to query Hasura.
  Following are the default files.
  ## config.json

  {
  	"HASURA_HOST": "http://localhost:8080",
  	"MQTT_HOST": "http://127.0.0.1:1883",
  	"MODE": "JSON",// change it to spBv1.0 if you want to decode using sparkplug b spec
  	"MQTT_CHANNEL": "payloads/#"
  }
  ## parse.js
  export function getMutation(message, topic) {
  	const device_id = topic.split('/')[1]
  	const timestamp = new Date().toISOString()
  	// console.log(message)
  	return {
  		query: 'mutation AddDeviceData($data: jsonb!, $device_id: String!, $timestamp: 
  timestamptz!){insert_device_data_one(object: {data: $data, device_id: $device_id,  timestamp:$timestamp}) {id}}',
  		variables: {
  			data: message,
  			device_id,
  			timestamp: message.timestamp
  				? new Date(message.timestamp).toISOString()
  				: timestamp,
  		},
  		operationName: 'AddDeviceData',
  	}
  }

See code: src/commands/init.js

Configuring Hasura GraphQL Engine.

Note : The following setup is indented to give a basic idea of setting up Hasura Engine for accepting data ingestions, you may design your tables/mutations as per your business logic

Setting up the DB

This project currently support only one Mutation (insert_device_data). To setup the table

• Go to Hasura Console > Data > SQL
• Run & Track the following SQL to create a table and setup the mutation

CREATE TABLE "public"."device_data"("id" serial NOT NULL, "data" jsonb NOT NULL, "timestamp" timestamptz NOT NULL, "device_id" text NOT NULL, PRIMARY KEY ("id") );

Authorization

Considering we use mutations, set of headers we pass is as follows

      "X-Hasura-Role": "device",
      "X-Hasura-User-Id": <device_id>,

device_id: Used here is directly parsed from the sparkplug topic.

So any custom permissions can be set on the hasura cloud with a role device. We recommend to use any of the following approaches.

• Allow from all device
• Allow only from Registered devices

Allow from all device

Following permission will allow hasura-connect to make mutation from any device_id.

Allow only from Registered devices

For doing this, we need to create one more table to store the information of registered devices. This will ensure that Hasura allows the mutations only from the registered devices.

• Run & Track the following SQL

CREATE TABLE "public"."devices"("id" serial NOT NULL, "device_id" text NOT NULL, "active" boolean NOT NULL DEFAULT true, PRIMARY KEY ("id") );

• Add device information to the newly created table.
• Setup Permissions for device_data insert.

{
	"_exists": {
		"_table": { "schema": "public", "name": "devices" },
		"_where": {
			"_and": [
				{ "device_id": { "_eq": "X-Hasura-User-Id" } },
				{ "active": { "_eq": true } }
			]
		}
	}
}

Above permission will make sure that the mutation is only allowed if the device is registered in the devices table.

Roadmap

• More configurations
• Support other streams (Apache Kafka, Apache NiFi, Other MQs, SNS, GCM etc.)
• Batch mutaions for better performance
• support custom MQTT payload spec
• make a Demo app (a digital twin/ realtime chart :?)
• Performance test & optimisation
• Dockerization
• CI implementaion