Skip to content
/ mist Public

A distributed, tag-based pub-sub service for modern web applications and container-driven cloud.

License

Notifications You must be signed in to change notification settings

nanopack/mist

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Mist

Build Status

Mist is a simple pub/sub based on the idea that messages are tagged. To subscribe, a client simply constructs a list of tags that it is interested in, and all messages that are tagged with all of those tags are sent to that client.

A client can not only be a subscriber (with multiple active subscriptions), but also a publisher. Clients will receive messages for any tags they are subscribed, except message publish by themselves.

Available Commands

Mist comes with two sets of available commands out of the box. Basic commands and Admin commands. It also has the ability to accept custom commands and handlers.

You can connect to mist with something like netcat; once connected you can simply type commands:

>> nc 127.0.0.1 1445
{"command":"publish", "tags":["hello"], "data":"world!"}

Basic Commands

Basic command are what provide the core functionality of mist. They allow you subscribe to and publish messages, see all of your active subscriptions and also unsubscribe from any tags you no longer want to receive messages for.

Command Description Example
auth authenticate with the server to enter "admin" mode {"command":"auth", "data":"TOKEN"}
ping ping the server to test for an active connection {"command":"ping"}
subscribe subscribe to messages for all tags in a group {"command":"subscribe", "tags":["hello"]}
unsubscribe unsubscribe from exact tags provided {"command":"unsubscribe", "tags":["hello"]}
publish publish data to the list of tags {"command":"publish", "tags":["hello"], "data":"world!"}
list list all active subscriptions for client {"command":"list"}

Admin Commands

If mist is started with an authenticator and a token then a client has the chance to validate that token on connect. Once validated mist adds some additional admin commands that allow the creation of token/tag combos that provide a layer of authentication when using basic commands.

Command Description Example
register register a token with a set of tags {"command":"register", "tags":["hello"], "data":"TOKEN"}
unregister removes a token from mist completely {"command":"unregister", "data":"TOKEN"}
set adds a set of tags to a token {"command":"set", "tags":["hello"], "data":"TOKEN"}
unset removes a set of tags from a token {"command":"unset", "tags":["hello"], "data":"TOKEN"}
tags show tags that are associated with a token {"command":"tags", "data":"TOKEN"}

Messages

All communications within mist are sent and received as JSON encoded/decoded messages:

Message struct {
  Command string   `json:"command"`
  Tags    []string `json:"tags"`
  Data    string   `json:"data,omitempty"`
  Error   string   `json:"error,omitempty"`
}

Each Message has a set of tags and data. Tags can take any form you like, as they are just an array of strings.

{
  "tags": ["company:pagodabox", "product:mist", "repo:#nanopack"],
  "data": "Mist is awesome!"
}

Subscribing / Publishing

Think of tags as a way to filter out messages you don't want to receive; the more tags that are added to a subscription the more direct a message has to be:

Subscribed tags Messages received from tags
["onefish"] ["onefish"], ["onefish","twofish"], ["onefish","twofish","redfish"]
["onefish", "twofish"] ["onefish","twofish"], ["onefish","twofish","redfish"]
["onefish", "twofish", "redfish"] ["onefish","twofish","redfish"]

Message that are published to clients as the result of a subscription are delivered in this format:

{"command":"<command>", "tags":["<tag>", "<tag>"], "data":"<data>"}

A few things to not about how mist handles data:

  • Data flowing through mist is not touched or verified in anyway, however, it MUST NOT contain a newline character as this will break the mist protocol.

  • Messages are not guaranteed to be delivered, if the client is running behind on processing messages, newer messages could be dropped.

  • Messages are not stored, if no client is available to receive the message, then it is dropped.

Listeners

Out of the box mist supports three different types of servers (TCP, HTTP, and Websocket). By default, when mist starts, it will start one of each.

TCP server listening at '127.0.0.1:1445'...
HTTP server listening at '127.0.0.1:8080'...
WS server listening at '127.0.0.1:8888'...

When starting mist, you can specify any number and type of server you'd like as long as it follows the string URI protocol (If a listener is passed that mist doesn't support it will skip).

Also, if mist doesn't support a server you need it allows you to register custom servers that can be used on startup.

Available listeners:

(scheme:[//[user:pass@]host[:port]][/]path[?query][#fragment])

Listener URI scheme
tcp tcp://127.0.0.1:1445
http http://127.0.0.1:8080
websocket ws://127.0.0.1:8888
Example
./mist --server --listeners "tcp://127.0.0.1:1445", "http://127.0.0.1:8080", "ws://127.0.0.1:8888"

Authenticators

Mist also provides support for authentication. This means that during startup you can provide mist with an authenticator and a token. Once enabled, any client that connects to the server has an opportunity (as the first command) to provide the authentication token to "unlock" admin commands for that connection.

./mist --server --authenticator memory:// --token TOKEN

Notice that when an authenticator is provided, a token must also be provided otherwise the server will not start.

Available authenticators

Like listeners, mist allows for the registration of custom authenticators.

(scheme:[//[user:pass@]host[:port]][/]path[?query][#fragment])

Authenticator URI scheme description
memory memory:// an in memory store
scribble scribble://?db=/tmp a tiny JSON database
postgres postgres://postgres@127.0.0.1:5432?db=postgres n/a

Connecting to an authenticated server

When connecting to an authenticated server, to enter "admin" mode, the very first communication across the wire must be the auth command.

>> nc 127.0.0.1 1445
{"command":"auth", "data":"TOKEN"}

Failing to authenticate will still allow the connection to proceed, however no "admin" commands will be allowed on the connection.

Clients:

Out of the box mist provides a CLI, a TCP client, and the ability to connect via Websocket Clients

CLI:

The CLI provides a built-in TCP client that can be used to connect with any running mist

Usage:
  mist [flags]
  mist [command]

Available Commands:
  ping        Ping a running mist server
  subscribe   Subscribe tags
  publish     Publish a message

Flags:
      --authenticator string   Setting enables authentication, storing tokens in the authenticator provided
      --config string          Path to config file
      --listeners value        A comma delimited list of servers to start (default [tcp://127.0.0.1:1445,ws://127.0.0.1:8888])
      --log-level string       Output level of logs (TRACE, DEBUG, INFO, WARN, ERROR, FATAL) (default "INFO")
      --server                 Run mist as a server
      --token string           Auth token for connections
  -v, --version                Display the current version of this CLI

Use "mist [command] --help" for more information about a command.

TCP Client

Here is an example of a TCP client written in Golang that will connect with mist and handle messages:

package main

import (
	"fmt"
	"os"
	"time"

	"github.com/nanopack/mist/clients"
)

func main() {
	client, err := clients.New("127.0.0.1:1445")
	if err != nil {
		os.Exit(1)
	}

	// example commands (not handling errors for brevity)
	client.Ping()
	client.Subscribe([]string{"hello"})
	client.Publish([]string{"hello"}, "world")
	client.List()
	// client.Unsubscribe([]string{"hello"})

	// do stuff with messages
	for {
		select {
		case msg := <-client.Messages():
			fmt.Printf("MSG: %#v\n", msg)
		case <-time.After(time.Second * 1):
			// do something if messages are taking too long
		}
	}

	// do stuff with messages (alternate)
	// for msg := range client.Messages() {
	// 	fmt.Printf("MSG: %#v\n", msg)
	// }
}

Websocket Client

Since mist just uses a JSON message protocol internally, sending messages via websocket is easy.

NOTE: If authentication is enabled you'll need to provide a token when connecting the websocket:

  • As a Header: X-AUTH-TOKEN: token
  • As a query param: x-auth-token=token

NOTE: If using WSS you'll need to accept the generated cert via the browser https://localhost:8082/subscribe/websocket prior to chrome console letting you run the following javascript.

Example
  // connect the websocket
  var ws = new WebSocket("ws://localhost:8080/subscribe/websocket?x-auth-token=token")

  // handle responses from the server
  ws.onmessage = function(me){
    console.log("Response!", me.data)
  }

  // ping
  ws.send(JSON.stringify({"command": "ping"}))

  // subscribe
  ws.send(JSON.stringify({"command": "subscribe", "tags": ["hello", "world"]}))

  // unsubscribe
  ws.send(JSON.stringify({"command": "unsubscribe", "tags": ["hello", "world"]}))

  // list
  ws.send(JSON.stringify({"command": "list"}))

Running mist:

To run mist as a server, using the following command will start mist as a daemon:

mist --server

If you need to override any default config options you can pass the path to a config file:

mist --config /path/to/config

example config
authenticator: memory://
listeners:
  - tcp://127.0.0.1:1445
log-level: INFO
token: TOKEN
server: true

Or you can just pass any configuration options as flags:

mist --server --log-level DEBUG

Contributing

Contributions to mist are welcome and encouraged. Mist is a Nanobox project and contributions should follow the Nanobox Contribution Process & Guidelines.

open source