4

This is my node js backend API method.

apiRouter.route('/makeComment')
        .post((req, res) => {

            consoleLogger.info("Inside makeComment API..");
            logger.info("Inside makeComment");

            let timestamp = new Date().getTime();

            let latestCommentID = timestamp.toString();

            console.log("comment ID generated is "+latestCommentID);


            res.json({
                                'makeComment': 'success',
                                'commentid':latestCommentID
                 });

        });

Now, if multiple concurrent requests come to this API, what will happen?

As per my understanding, NodeJS will maintain a Event Queue for the requests and process requests one by one.

So, it is impossible to get the same timestamp for two different requests.

Please let me know if my understanding is correct?

Edit 1:

After googling for some time, got this link, which clearly explains some of the concepts.

DukeLover
  • 2,184
  • 2
  • 19
  • 27
  • @SaurabhMistry : How? If no multithreading happening here, then how is it possible? Do you mean to say NodeJS parallelizes the requests ? – DukeLover Jun 06 '19 at 05:16
  • @SaurabhMistry : I am not making any async or parallel call here, then how is it possible ? – DukeLover Jun 06 '19 at 05:23

3 Answers3

7

You can absolutely end up with the same timestamp on 2 concurrent calls. But not because NodeJS or your server served the request in parallel, but rather because a millisecond is a long enough time that your server could process many requests in the same millisecond.

Multi-threaded vs. Concurrent

You have correctly identified a subtle but important distinction between concurrency and multi-threading. In a multi-threaded environment, two different operations can truly take place at the same time on 2 different cores of a CPU. In a concurrent but single-threaded environment, things do happen sequentially, hopefully in a non-blocking manner, and really fast!

Javascript is a single threaded environment. And yes, it has an event loop where it queues tasks and processes them one at a time. NodeJS (and Web APIs in case of browser Javascript apps) offer certain async functions, such as fs.writeFile or fetch, which the runtime in collaboration with the OS may perform in the same or different thread/core, and then orchestrate returning results back to your application via callbacks and Promises.

In case of your example, your handler (the part starting from (req, res) => { ... }) consists of sync calls only. So, yes, various calls to this handler will run sequentially, one after another. And, while no two runs of this handler will ever truly happen at the same time, but they will happen really fast and you will likely have cases where you'll get the same millisecond value from the Date object. If you had a higher resolution timestamp available (maybe nano seconds) or if your handler took longer to run (for example, you ran a for loop a billion iterations), then you'll be able to observe the sequential behavior more clearly.

Avoid blocking (sync) calls in single-threaded applications

This is the exact reason why you are advised against performing any synchronous IO operation (e.g. fs.writeFileSync, etc.) in a NodeJS web server, because while one request is performing a blocking IO operation, all other requests are waiting, queued on the event loop.

Really good video about Javascript event loop; this should illuminate some topics: https://www.youtube.com/watch?v=8aGhZQkoFbQ

Arash Motamedi
  • 9,284
  • 5
  • 34
  • 43
5

There is no possibility of concurrency here. As you have correctly studied, Node.js runs in a single-threaded environment (unless you use clusters, the worker API, or other related Node.js modules that facilitate IPC). As such, the line

let timestamp = new Date().getTime();

cannot be concurrently invoked by two simultaneously running threads, excluding the exceptions noted above.

However, Date.prototype.getTime() only has millisecond resolution, so it's remotely possible that the callback to apiRouter.route('/makeComment').post() might be sequentially invoked by the event loop from two pending asynchronous requests within the same millisecond.

Patrick Roberts
  • 49,224
  • 10
  • 102
  • 153
  • Hi Patrick : Thanks for the clarification. I was confused about my understanding. You saved me. :-) Now , in order to mitigate the remotely possible chance , if I take nanosecond resolution (as mentioned by @Arash Motamedi ) , will it solve the purpose ? – DukeLover Jun 06 '19 at 05:50
  • 1
    @DukeLover nanosecond resolution would certainly mitigate the issue, but what I'd suggest doing is persisting comments with a database, then using the id from the database entry's row or document as the comment id, rather than relying on a time-based mechanism to produce unique identifiers. – Patrick Roberts Jun 06 '19 at 05:52
  • Thanks a lot again. :-) . I will definitely keep this point in mind. – DukeLover Jun 06 '19 at 05:57
  • 1
    @DukeLover if your goal is to generate an ID, either do as Patrick has suggested (write to a DB that generates a unique ID), or use a UUID library, or even simpler, create a variable outside your handler `let i = 0` and increment and use it inside your handler `let id = ++i`. – Arash Motamedi Jun 06 '19 at 18:46
0

It is correct, that JavaScript runs in a single-thread environment thus 2 queries can not be invoked simultaneously. However, time precision is essential, as less precise time is being measured more likely that such a small query would fall into the same fraction of time.

I'm talking from the physics perspective, so it all depends on the precession of the system.

Scene
  • 137
  • 1
  • 5