lib: fix cluster handle leak

It is possible to cause a resource leak in SharedHandle. This commit
fixes the leak.

Fixes: #2510
PR-URL: #5152
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: James M Snell <jasnell@gmail.com>

lib/cluster.js

@@ -363,7 +363,10 @@ function masterInit() {
        * if it has disconnected, otherwise we might
        * still want to access it.
        */
-      if (!worker.isConnected()) removeWorker(worker);
+      if (!worker.isConnected()) {
+        removeHandlesForWorker(worker);
+        removeWorker(worker);
+      }
 
       worker.suicide = !!worker.suicide;
       worker.state = 'dead';

test/simple/test-cluster-shared-leak.js

@@ -0,0 +1,48 @@
+// On some platforms this test triggers an assertion in cluster.js.
+// The assertion protects against memory leaks.
+// https://github.com/nodejs/node/pull/3510
+
+'use strict';
+var common = require('../common');
+var assert = require('assert');
+var net = require('net');
+var cluster = require('cluster');
+cluster.schedulingPolicy = cluster.SCHED_NONE;
+
+if (cluster.isMaster) {
+  var conn, worker1, worker2;
+
+  worker1 = cluster.fork();
+  worker1.on('message', common.mustCall(function() {
+    worker2 = cluster.fork();
+    worker2.on('online', function() {
+      conn = net.connect(common.PORT, common.mustCall(function() {
+        worker1.disconnect();
+        worker2.disconnect();
+      }));
+      conn.on('error', function(e) {
+        // ECONNRESET is OK
+        if (e.code !== 'ECONNRESET')
+          throw e;
+      });
+    });
+  }));
+
+  cluster.on('exit', function(worker, exitCode, signalCode) {
+    assert(worker === worker1 || worker === worker2);
+    assert.strictEqual(exitCode, 0);
+    assert.strictEqual(signalCode, null);
+    if (Object.keys(cluster.workers).length === 0)
+      conn.destroy();
+  });
+
+  return;
+}
+
+var server = net.createServer(function(c) {
+  c.end('bye');
+});
+
+server.listen(common.PORT, function() {
+  process.send('listening');
+});