seneca-gossip

seneca-gossip

seneca-gossip a cluster solution using gossip protocols for seneca.

Feature

• Cluster : A set of nodes joined together through the membership service.
• Auto-discovery : Auto make client when a new node joined.

Usage

The gossip's port default is options.port+1 .

The server2 join the cluster with seed - 127.0.0.1:4445.

The client join the cluster with seed - 127.0.0.1:4445 too, it also can client to the server2 with gossip.

So you can start a node with a known active node in the cluster, and then you can call the service with {role: xxx, cmd:xxx} but don't need to care about their physical location in the cluster.

Server

var Seneca = require('seneca');
var Gossip = new require('../index');

var seneca = new Seneca();
seneca.add({role: 'api', cmd: 'add'}, function (msg, respond) {
    var result = msg.left + msg.right;
    respond(null, {result: result});
});

var options = {port: 4444, type: 'tcp', host: '127.0.0.1', pin: 'role:api'};

Gossip(seneca).start(options);


var seneca2 = new Seneca();
seneca2.add({role: 'api', cmd: 'mcl'}, function (msg, respond) {
    var result = msg.left * msg.right;
    respond(null, {result: result});
});

var options2 = {port: 4454, type: 'tcp', host: '127.0.0.1', pin: 'role:api',
    seeds: ['127.0.0.1:4445']};

Gossip(seneca2).start(options2);

Client

var Seneca = require('seneca');

var seneca = new Seneca();
var Gossip = new require('../index');

Gossip(seneca).start({port: 6666, type: 'tcp', host: '127.0.0.1', seeds: ['127.0.0.1:4445']});


setInterval(function () {
    seneca.act({role: "api", cmd: "add", left: 2, right: 3}, function (e, r) {
        console.log("2 + 3 =", r.result);
    });

    seneca.act({role: "api", cmd: "mcl", left: 2, right: 3}, function (e, r) {
        console.log("2 * 3 =", r.result);
    });
}, 2000);

**************************
2 + 3 = 5
2 * 3 = 6

TODO

• add test cases
• handle some exception. like outtime...
• add cluster sharding support
• ...

Acknowledgements

Both the gossip protocol and the failure detection algorithms are based off of academic papers and Cassandra's (http://www.cassandra.org/) implementation of those papers. This library is highly indebted to both.

• "Efficient reconciliation and flow control for anti-entropy protocols"
• "The Phi accrual failure detector"