下载大型媒体文件时,请求会使进程内存不足

Question

我写了一个简单的脚本来从CDN下载视频文件,其中直接URL很容易生成,例如http://something.com/N.mp4,其中N是数字.

问题是,当下载大于300MB的文件时,文件完全出现在硬盘中,但在request(...)回调之前,会发生内存分配失败:

FATAL ERROR: CALL_AND_RETRY_0 Allocation failed - process out of memory

这是否因为一些严重的不良做法而发生？可以request下载这个尺寸的媒体文件吗？

环境:Win7,4GB +可用RAM,节点v0.10.31

var request = require('request');
var async = require('async');
var fs = require('fs');
var start = +process.argv[2] || 1;
var end = +process.argv[3] || 50;
var url = 'http://something.com/';

try {
  fs.mkdirSync(__dirname + '/videos/');
} catch (e) {}

var index = start;

async.whilst(
  function () { return index <= end; },
  function (callback) {
    var fileName = index + '.mp4';
    console.log('Started: ' + fileName);
    console.time('Done (' + fileName + ')');
    request(url + fileName, function() {
      console.timeEnd('Done (' + fileName + ')');
      index++;
      callback(null);
    }).pipe(fs.createWriteStream(__dirname + '/videos/' + fileName));
  },
  function (err) {
    if (err) {
      return console.error(err);
    }
    console.log('Script finished.');
  }
);

控制台输出示例:

> node index.js 3
Started: 3.mp4
Done (3.mp4): 296592ms
Started: 4.mp4
Done (4.mp4): 369718ms
Started: 5.mp4
FATAL ERROR: CALL_AND_RETRY_0 Allocation failed - process out of memory

Answer 1

如果您使用request带回调的模块,它会将整个响应主体缓冲在内存中.尝试省略回调并使用finishfs流的事件代替.

var writer = fs.createWriteStream(__dirname + '/videos/' + fileName);
writer.on('finish', function() {
  // ...
  index++;
  callback(null);
});
request(url + fileName).pipe(writer);