A single instance of Node runs in a single thread. To take advantage of\nmulti-core systems the user will sometimes want to launch a cluster of Node\nprocesses to handle the load.\n\n
\nThe cluster module allows you to easily create a network of processes that\nall share server ports.\n\n
\nvar cluster = require('cluster');\nvar http = require('http');\nvar numCPUs = require('os').cpus().length;\n\nif (cluster.isMaster) {\n // Fork workers.\n for (var i = 0; i < numCPUs; i++) {\n cluster.fork();\n }\n\n cluster.on('exit', function(worker, code, signal) {\n console.log('worker ' + worker.process.pid + ' died');\n });\n} else {\n // Workers can share any TCP connection\n // In this case its a HTTP server\n http.createServer(function(req, res) {\n res.writeHead(200);\n res.end("hello world\\n");\n }).listen(8000);\n}Running node will now share port 8000 between the workers:\n\n
\n% NODE_DEBUG=cluster node server.js\n23521,Master Worker 23524 online\n23521,Master Worker 23526 online\n23521,Master Worker 23523 online\n23521,Master Worker 23528 onlineThis feature was introduced recently, and may change in future versions.\nPlease try it out and provide feedback.\n\n
\nAlso note that, on Windows, it is not yet possible to set up a named pipe\nserver in a worker.\n\n
\n", "miscs": [ { "textRaw": "How It Works", "name": "How It Works", "type": "misc", "desc": "The worker processes are spawned using the child_process.fork method,\nso that they can communicate with the parent via IPC and pass server\nhandles back and forth.\n\n
The cluster module supports two methods of distributing incoming\nconnections.\n\n
\nThe first one (and the default one on all platforms except Windows),\nis the round-robin approach, where the master process listens on a\nport, accepts new connections and distributes them across the workers\nin a round-robin fashion, with some built-in smarts to avoid\noverloading a worker process.\n\n
\nThe second approach is where the master process creates the listen\nsocket and sends it to interested workers. The workers then accept\nincoming connections directly.\n\n
\nThe second approach should, in theory, give the best performance.\nIn practice however, distribution tends to be very unbalanced due\nto operating system scheduler vagaries. Loads have been observed\nwhere over 70% of all connections ended up in just two processes,\nout of a total of eight.\n\n
\nBecause server.listen() hands off most of the work to the master\nprocess, there are three cases where the behavior between a normal\nnode.js process and a cluster worker differs:\n\n
server.listen({fd: 7}) Because the message is passed to the master,\nfile descriptor 7 in the parent will be listened on, and the\nhandle passed to the worker, rather than listening to the worker's\nidea of what the number 7 file descriptor references.server.listen(handle) Listening on handles explicitly will cause\nthe worker to use the supplied handle, rather than talk to the master\nprocess. If the worker already has the handle, then it's presumed\nthat you know what you are doing.server.listen(0) Normally, this will cause servers to listen on a\nrandom port. However, in a cluster, each worker will receive the\nsame "random" port each time they do listen(0). In essence, the\nport is random the first time, but predictable thereafter. If you\nwant to listen on a unique port, generate a port number based on the\ncluster worker ID.There is no routing logic in Node.js, or in your program, and no shared\nstate between the workers. Therefore, it is important to design your\nprogram such that it does not rely too heavily on in-memory data objects\nfor things like sessions and login.\n\n
\nBecause workers are all separate processes, they can be killed or\nre-spawned depending on your program's needs, without affecting other\nworkers. As long as there are some workers still alive, the server will\ncontinue to accept connections. Node does not automatically manage the\nnumber of workers for you, however. It is your responsibility to manage\nthe worker pool for your application's needs.\n\n
\n" } ], "properties": [ { "textRaw": "cluster.schedulingPolicy", "name": "schedulingPolicy", "desc": "The scheduling policy, either cluster.SCHED_RR for round-robin or\ncluster.SCHED_NONE to leave it to the operating system. This is a\nglobal setting and effectively frozen once you spawn the first worker\nor call cluster.setupMaster(), whatever comes first.\n\n
SCHED_RR is the default on all operating systems except Windows.\nWindows will change to SCHED_RR once libuv is able to effectively\ndistribute IOCP handles without incurring a large performance hit.\n\n
cluster.schedulingPolicy can also be set through the\nNODE_CLUSTER_SCHED_POLICY environment variable. Valid\nvalues are "rr" and "none".\n\n
All settings set by the .setupMaster is stored in this settings object.\nThis object is not supposed to be changed or set manually, by you.\n\n
True if the process is a master. This is determined\nby the process.env.NODE_UNIQUE_ID. If process.env.NODE_UNIQUE_ID is\nundefined, then isMaster is true.\n\n
This boolean flag is true if the process is a worker forked from a master.\nIf the process.env.NODE_UNIQUE_ID is set to a value, then\nisWorker is true.\n\n
A reference to the current worker object. Not available in the master process.\n\n
\nvar cluster = require('cluster');\n\nif (cluster.isMaster) {\n console.log('I am master');\n cluster.fork();\n cluster.fork();\n} else if (cluster.isWorker) {\n console.log('I am worker #' + cluster.worker.id);\n}A hash that stores the active worker objects, keyed by id field. Makes it\neasy to loop through all the workers. It is only available in the master\nprocess.\n\n
// Go through all workers\nfunction eachWorker(callback) {\n for (var id in cluster.workers) {\n callback(cluster.workers[id]);\n }\n}\neachWorker(function(worker) {\n worker.send('big announcement to all workers');\n});Should you wish to reference a worker over a communication channel, using\nthe worker's unique id is the easiest way to find the worker.\n\n
\nsocket.on('data', function(id) {\n var worker = cluster.workers[id];\n});When a new worker is forked the cluster module will emit a 'fork' event.\nThis can be used to log worker activity, and create you own timeout.\n\n
\nvar timeouts = [];\nfunction errorMsg() {\n console.error("Something must be wrong with the connection ...");\n}\n\ncluster.on('fork', function(worker) {\n timeouts[worker.id] = setTimeout(errorMsg, 2000);\n});\ncluster.on('listening', function(worker, address) {\n clearTimeout(timeouts[worker.id]);\n});\ncluster.on('exit', function(worker, code, signal) {\n clearTimeout(timeouts[worker.id]);\n errorMsg();\n});After forking a new worker, the worker should respond with a online message.\nWhen the master receives a online message it will emit such event.\nThe difference between 'fork' and 'online' is that fork is emitted when the\nmaster tries to fork a worker, and 'online' is emitted when the worker is\nbeing executed.\n\n
\ncluster.on('online', function(worker) {\n console.log("Yay, the worker responded after it was forked");\n});When calling listen() from a worker, a 'listening' event is automatically assigned\nto the server instance. When the server is listening a message is send to the master\nwhere the 'listening' event is emitted.\n\n
The event handler is executed with two arguments, the worker contains the worker\nobject and the address object contains the following connection properties:\naddress, port and addressType. This is very useful if the worker is listening\non more than one address.\n\n
cluster.on('listening', function(worker, address) {\n console.log("A worker is now connected to " + address.address + ":" + address.port);\n});When a workers IPC channel has disconnected this event is emitted.\nThis will happen when the worker dies, usually after calling\n.kill().\n\n
When calling .disconnect(), there may be a delay between the\ndisconnect and exit events. This event can be used to detect if\nthe process is stuck in a cleanup or if there are long-living\nconnections.\n\n
cluster.on('disconnect', function(worker) {\n console.log('The worker #' + worker.id + ' has disconnected');\n});When any of the workers die the cluster module will emit the 'exit' event.\nThis can be used to restart the worker by calling fork() again.\n\n
cluster.on('exit', function(worker, code, signal) {\n var exitCode = worker.process.exitCode;\n console.log('worker ' + worker.process.pid + ' died ('+exitCode+'). restarting...');\n cluster.fork();\n});When the .setupMaster() function has been executed this event emits.\nIf .setupMaster() was not executed before fork() this function will\ncall .setupMaster() with no arguments.\n\n
setupMaster is used to change the default 'fork' behavior. The new settings\nare effective immediately and permanently, they cannot be changed later on.\n\n
Example:\n\n
\nvar cluster = require("cluster");\ncluster.setupMaster({\n exec : "worker.js",\n args : ["--use", "https"],\n silent : true\n});\ncluster.fork();Spawn a new worker process. This can only be called from the master process.\n\n
\n" }, { "textRaw": "cluster.disconnect([callback])", "type": "method", "name": "disconnect", "signatures": [ { "params": [ { "textRaw": "`callback` {Function} called when all workers are disconnected and handlers are closed ", "name": "callback", "type": "Function", "desc": "called when all workers are disconnected and handlers are closed", "optional": true } ] }, { "params": [ { "name": "callback", "optional": true } ] } ], "desc": "When calling this method, all workers will commit a graceful suicide. When they are\ndisconnected all internal handlers will be closed, allowing the master process to\ndie graceful if no other event is waiting.\n\n
\nThe method takes an optional callback argument which will be called when finished.\n\n
\n" } ], "classes": [ { "textRaw": "Class: Worker", "type": "class", "name": "Worker", "desc": "A Worker object contains all public information and method about a worker.\nIn the master it can be obtained using cluster.workers. In a worker\nit can be obtained using cluster.worker.\n\n
Each new worker is given its own unique id, this id is stored in the\nid.\n\n
While a worker is alive, this is the key that indexes it in\ncluster.workers\n\n
\n" }, { "textRaw": "`process` {ChildProcess object} ", "name": "process", "desc": "All workers are created using child_process.fork(), the returned object\nfrom this function is stored in process.\n\n
See: Child Process module\n\n
\n" }, { "textRaw": "`suicide` {Boolean} ", "name": "suicide", "desc": "This property is a boolean. It is set when a worker dies after calling\n.kill() or immediately after calling the .disconnect() method.\nUntil then it is undefined.\n\n
This function is equal to the send methods provided by\nchild_process.fork(). In the master you should use this function to\nsend a message to a specific worker. However in a worker you can also use\nprocess.send(message), since this is the same function.\n\n
This example will echo back all messages from the master:\n\n
\nif (cluster.isMaster) {\n var worker = cluster.fork();\n worker.send('hi there');\n\n} else if (cluster.isWorker) {\n process.on('message', function(msg) {\n process.send(msg);\n });\n}This function will kill the worker, and inform the master to not spawn a\nnew worker. The boolean suicide lets you distinguish between voluntary\nand accidental exit.\n\n
cluster.on('exit', function(worker, code, signal) {\n if (worker.suicide === true) {\n console.log('Oh, it was just suicide\\' – no need to worry').\n }\n});\n\n// kill worker\nworker.kill();This method is aliased as worker.destroy() for backwards\ncompatibility.\n\n
When calling this function the worker will no longer accept new connections, but\nthey will be handled by any other listening worker. Existing connection will be\nallowed to exit as usual. When no more connections exist, the IPC channel to the worker\nwill close allowing it to die graceful. When the IPC channel is closed the disconnect\nevent will emit, this is then followed by the exit event, there is emitted when\nthe worker finally die.\n\n
Because there might be long living connections, it is useful to implement a timeout.\nThis example ask the worker to disconnect and after 2 seconds it will destroy the\nserver. An alternative would be to execute worker.kill() after 2 seconds, but\nthat would normally not allow the worker to do any cleanup if needed.\n\n
if (cluster.isMaster) {\n var worker = cluster.fork();\n var timeout;\n\n worker.on('listening', function(address) {\n worker.disconnect();\n timeout = setTimeout(function() {\n worker.send('force kill');\n }, 2000);\n });\n\n worker.on('disconnect', function() {\n clearTimeout(timeout);\n });\n\n} else if (cluster.isWorker) {\n var net = require('net');\n var server = net.createServer(function(socket) {\n // connection never end\n });\n\n server.listen(8000);\n\n server.on('close', function() {\n // cleanup\n });\n\n process.on('message', function(msg) {\n if (msg === 'force kill') {\n server.close();\n }\n });\n}This event is the same as the one provided by child_process.fork().\nIn the master you should use this event, however in a worker you can also use\nprocess.on('message')\n\n
As an example, here is a cluster that keeps count of the number of requests\nin the master process using the message system:\n\n
\nvar cluster = require('cluster');\nvar http = require('http');\n\nif (cluster.isMaster) {\n\n // Keep track of http requests\n var numReqs = 0;\n setInterval(function() {\n console.log("numReqs =", numReqs);\n }, 1000);\n\n // Count requestes\n function messageHandler(msg) {\n if (msg.cmd && msg.cmd == 'notifyRequest') {\n numReqs += 1;\n }\n }\n\n // Start workers and listen for messages containing notifyRequest\n var numCPUs = require('os').cpus().length;\n for (var i = 0; i < numCPUs; i++) {\n cluster.fork();\n }\n\n Object.keys(cluster.workers).forEach(function(id) {\n cluster.workers[id].on('message', messageHandler);\n });\n\n} else {\n\n // Worker processes have a http server.\n http.Server(function(req, res) {\n res.writeHead(200);\n res.end("hello world\\n");\n\n // notify master about the request\n process.send({ cmd: 'notifyRequest' });\n }).listen(8000);\n}Same as the cluster.on('online') event, but emits only when the state change\non the specified worker.\n\n
cluster.fork().on('online', function() {\n // Worker is online\n});Same as the cluster.on('listening') event, but emits only when the state change\non the specified worker.\n\n
cluster.fork().on('listening', function(address) {\n // Worker is listening\n});Same as the cluster.on('disconnect') event, but emits only when the state change\non the specified worker.\n\n
cluster.fork().on('disconnect', function() {\n // Worker has disconnected\n});Emitted by the individual worker instance, when the underlying child process\nis terminated. See child_process event: 'exit'.\n\n
\nvar worker = cluster.fork();\nworker.on('exit', function(code, signal) {\n if( signal ) {\n console.log("worker was killed by signal: "+signal);\n } else if( code !== 0 ) {\n console.log("worker exited with error code: "+code);\n } else {\n console.log("worker success!");\n }\n});