cursor returned from find takes minutes to start iterating over result set

[Allan Edwards]

Hi everyone... I am running a 3.0.1 version of mongodb on Ubuntu Linux.  I have a node.js web application that pulls some data from the server.  What is weird is I run the find command and immediatly get a cursor back with no errors.  Then when I call any method on the cursor that is returned it blocks for a very long time before returning via the callback.  I am using the 2.x driver for node.js.

Here is some sample code...would anyone have any idea why this request is blocking for so long?

GLOBAL.db.collection("data").find({ dataId : dataId }, { _id : 1, data : 2 }, { "sort" : [[ '_id', 1 ]] }, function(err, docscursor) {
    if (err) {
        cb(err);
    }
    else {
        docscursor.batchSize(1, function(err, thisobject) {
            var doc = null;

            async.doWhilst(
                function(callback) {

                    docscursor.next(function (err, adoc) {
                        doc = adoc;

                        async.series([
                                function(callback){
                                    self.logTransfer(securityUserId, doc.data.length(), function (err) {
                                        callback(err);
                                    });
                                },
                                function(callback){
                                    res.write(new Buffer(doc.data))
                                    callback(null);
                                }
                            ],
                            function (err) {
                                callback(err);
                            }
                        );
                    });
                },
                function test() {
                    if (doc == null)
                        return false;
                    else
                        return true;
                },
                function(err) {
                    cb(err);
                });
        });
    }
});

Thanks!!!
Allan

[Christian Amor Kvalheim]

Hi Allan

Setting a batch size of 1 is always bad as it forces the driver to perform a roundtrip to the server for each document causing massive slowdown of the performance. This is a  situation where you can simplify your code by using streams instead the complexity of dealing with async.

https://gist.github.com/christkv/ae7cf17696cc2545ecbd