node-apn: Provider does not shutdown after a successful send

I am facing this problem too. I am simply using node test.js to run my script, and after getting the response, the process won’t exit immediately after calling shutdown().

I had to do something like this to exit the process, but not sure if this is the right way to handle.

connection.send(payload, deviceTokens)
    .then(res => {
        console.log(res); 
        // connection.shutdown(); // won't work
        connection.client.endpointManager._endpoints.forEach(endpoint => endpoint.destroy()); // will do
    })
    .catch(err => console.log(err))

Any help would be appreciated!

So what I did for this was to make a promise, which waits for the send to finish before firing.

            apnProvider.send(notification, deviceToken).then(function(notificationResult) {
                // Check the result for any failed devices
                console.log(notificationResult);
                apnProvider.shutdown();
            });

putting a .then here is a promise that is called once the the function is finished and gives me the result so I can use it in the next function.

This should properly shut down your apn.Provider after it sends.

I use @tri613 's solution and remove heart beat check

when using node-apn on aws lambda, I think _heartBeatIntervalCheck() can be removed.

AWS lambda is one time invocation and no need to keep connection, am I right?

I have the same issue, it can be fixed by adding the line clearInterval(this._heartBeatIntervalCheck); at the beginning of the method Endpoint.prototype.close in the file lib/protocol/endpoint.js.

So the method will be:

Endpoint.prototype.close = function close() {
    clearInterval(this._heartBeatIntervalCheck);
    if (this._acquiredStreamSlots === 0) {
      this._connection.close();
    }
    this._closePending = true;
  };

I think it’s obvious because the same line is in the destroy method. So if we clear the interval in the destroy method why don’t we clear it in the close method?

I hope somebody creates a pull request.

Has there been any update on this? This is absolutely killing the ability to use this with AWS Lambda.

Hi, I’m having exactly the same issue as described by @DJHoltkamp, when a successful notification is sent the node process doesn’t complete when provider.shutdown() is called. However, if only a failed notification is sent then the process does complete when provider.shutdown() is called. The process finally completes after about 60s, causing timeouts on AWS Lambda, and I’ve experienced this on Lambda using node V4.3 and on my local environment using node V4.5

I’ve been using the promise as @DRSisco described, but still having this issue…

    var apnProvider = new apn.Provider(APN_OPTIONS);
    var notification = new apn.Notification({
        alert: PUSH_BODY,
        sound: "default",
        contentAvailable: 1,
        payload: {
            "sender": "node-apn"
        },
        topic: PUSH_TOPIC
    });

    console.log(deviceTokens);
    apnProvider.send(notification, deviceTokens).then(function(response){
        console.log(JSON.stringify(response));
        apnProvider.shutdown();
        callback();
    });

I’m looking into the library to see if I can find the problem, tried @DJHoltkamp solution with heartBeatIntervalCheck, but no luck so far, any other suggestions?

Ok, so possibly a better solution is to stop the timer after the connection closes

  Endpoint.prototype.close = function close() {
    if (this._acquiredStreamSlots === 0) {
      clearInterval(this._heartBeatIntervalCheck);
      this._connection.close();
      console.log('closing connection');
    }
    this._closePending = true;
  };

As I said, I am not experienced with Node.js (or this library). Is this an ok change to make? One thing to note is that the Endpoint.destroy() never gets called under my use case now, but the process does finish quickly as it should