节点集群我如何只派生给定节点的 1 个实例？

Question

好的，所以我有一个用 Nodejs 编写的 API，它被集群化以利用使用“集群”模块的所有可用内核，总共有 4 个子进程。

const cluster = require('cluster');

我有一个进程（'remNodeProcess'），我只需要派生 1 个实例，即我不希望每个工作节点派生自己的 'remNodeProcess' 实例。这个“remNodeProcess”不需要对每个子节点都可用。为此，我正在尝试使用“子进程”模块。

const childProcess = require('child_process');

我对 Node Clusters 有很好的阅读，并且已经使用了一段时间。集群按预期工作，我得到了预期的结果，但我似乎无法弄清楚如何分叉一个进程的单个实例。我只想要集群中这个进程的 1 个实例。

这是集群的完整代码：

const setupWorkerThreadNodes = () => {

    // Get system CPUs
    let SYS_CORE_COUNT = require('os').cpus().length;

    // Read the Config Thread Configuration, if it exceeds the SYS_CORE_COUNT, set it to max usage, otherwise honour the Config
    let coreCount;
    if (webAppConfig.appConfig.nodeClusterConfiguration.mtMaxWorkerThreadCount) {
        coreCount = webAppConfig.appConfig.nodeClusterConfiguration.mtMaxWorkerThreadCount;
    } else if (webAppConfig.appConfig.nodeClusterConfiguration.mtMaxWorkerThreadCount > SYS_CORE_COUNT) {
        coreCount = SYS_CORE_COUNT;
    }

    logMsg(`NODE_MASTER PID: ${process.pid}`);
    console.log('NODE_CLUSTER Master is setting up Nodes... ');

    // For each Core, create a node and push it to the system cluster array
    for (let i = 0; i < coreCount; i++) {
        console.log(`NODE_CLUSTER Master is setting up Node [${i}/${coreCount}]... `);
        workers.push(cluster.fork({ WorkerName: `Worker${i}` }));

        // When a Worker responds to the Master, log the response
        workers[i].on('message', function (message) {
            console.log(message);
        });
    }

    // process is clustered on a core and process id is assigned
    cluster.on('online', function (worker) {
        console.log('Worker ' + worker.process.pid + ' is listening');
        logMsg(`CLUSTER_NODE Child opened. PID: ${worker.process.pid}`)
    });

    // If the Workers count matches the CoreCount, we've opened the correct amount of threads
    if (workers.length === coreCount) logMsg(`NODE_CLUSTER Array successfully established.`, 'info');

    // if any of the worker process dies then start a new one by simply forking another one
    cluster.on('exit', function (worker, code, signal) {
        console.log('Worker ' + worker.process.pid + ' died with code: ' + code + ', and signal: ' + signal);
        logMsg(`CLUSTER_NODE Child terminated. PID: ${worker.process.pid}. Code: ${code}. Signal: ${signal}`)
        console.log('Starting a new worker');
        cluster.fork();
        workers.push(cluster.fork());
        // to receive messages from worker process
        workers[workers.length - 1].on('message', function (message) {
            console.log(message);
        });
    });
};

我尝试了以下方法：

if(cluster.isMaster){
   const remNodeProcess= childProcess.fork('./background_workers/remNode.js');
   remNodeProcess.send(pool, (err) => { console.log(err) })
}

当我使用此代码运行 API 时，remNodeProcess 没有打开。如果我使用以下代码（没有 IF 包装器）：

const remNodeProcess= childProcess.fork('./background_workers/remNode.js');
remNodeProcess.send(pool, (err) => { console.log(err) })

我为每个孩子获得一个 remNodeProcess。

如何 fork 给定进程的一个实例？这可以从Master那里完成吗？

Answer 1

我想通了。

如果当前工作人员的 WorkerName 为 Worker0，这将手动分叉提醒子项。结果是集群中只有 1 个 remNode 实例，而不是 4 个。

if (process.env.WorkerName === "Worker0") {
   console.log(`${process.env.WorkerName} forking ReminderNode...`)
   remNodeProcess = childProcess.fork('./background_workers/remNode.js')
      .on('message', msg => {
         logMsg(msg)
      });
}