NodeJs Stream - Out of Memory

Question

I'm trying to process a stream of 300 million rows of data. Once I get to around 5 million rows I am getting a FATAL ERROR: CALL_AND_RETRY_LAST Allocation failed - process out of memory. (The number varies by machine but is consistent happening).

You can run the code below to see this happening - I can't tell if the issue in the code of inherent to streams. I tried to dumb the process but I couldn't do that.

Is there a memory limit? I removed all other code and 'dumbed' the example down to make sure it wasn't some back pressure issue.

    var Readable = require('stream').Readable;
    var Writable = require('stream').Writable;
    var util = require('util');
    var tenMillion = 10000000;
    //var tenMillion = 5000000; //THIS WORKS
    var writeEvery = tenMillion / 10;

    /*
    * Create a really simple stream that will run 10 million times 
    */
    function Streamo(max) {
        Readable.call(this, { objectMode: true });
        this._currentIndex = -1;
        this._maxIndex = max;
    }

    util.inherits(Streamo, Readable);

    Streamo.prototype._read = function () {
        this._currentIndex += 1;
        if (this._currentIndex % writeEvery == 0) {
            console.log(this._currentIndex + ' of ' + this._maxIndex)
        };

        if (this._currentIndex < 0 || this._currentIndex >= this._maxIndex) {
            console.log("BOOM")
            this.push(null);
            return;
        }
        this.push(true);
    };

    /*
    * Create a really simple Writable Stream to Count 
    */

    function Counta() {
        Writable.call(this, { objectMode: true, highWaterMark: (200 * 1024) });
        this._count = 0;
    }
    util.inherits(Counta, Writable);

    Counta.prototype._write = function (chunk, enc, cb) {
        this._count++;
        if (this._count % writeEvery == 0) {
            console.log('_______________________________' + this._count)
        };
        cb();
    };

    Counta.prototype.Count = function () {
        return this._count;
    }


    /*
    * Exercise It 
    */
    var s = new Streamo(tenMillion);
    var c = new Counta();
    s.pipe(c);
    c.on('finish', function () {
        console.log("BOOM BOOM BOOM BOOM BOOM BOOM BOOM BOOM BOOM ")
    });

Answer 1

This is a known issue with the current streams implementation.

In the streams documentation and code, there are multiple places where it is alluded to that _read() should be asynchronous.

So if you're not actually performing (async) i/o of some kind in your _read() implementation, then you may need to (at least occasionally) call setImmediate() before push()ing, to keep the call stack from getting too large. For example, this works without crashing:

Streamo.prototype._read = function (n) {
    this._currentIndex += 1;
    if (this._currentIndex % writeEvery == 0) {
        console.log(this._currentIndex + ' of ' + this._maxIndex)
    };

    if (this._currentIndex < 0 || this._currentIndex >= this._maxIndex) {
        console.log("BOOM")
        this.push(null);
        return;
    }
    var self = this;
    if (this._currentIndex % writeEvery == 0) {
      setImmediate(function() {
        self.push(true);
      });
    } else
      this.push(true);
};