Webhook subscription is triggering with no body

I encountered the same problem starting a few days ago. It turned out that the webhooks are now being sent with Transfer-Encoding: chunked, which means there is no content-length header. We were looking for content-length > 0 to see if there was a message body to read, otherwise assuming it was the initial validation request. I just had to change it to always read the content body -- the web framework will handle the transfer encoding, and we'll get back an empty string if there is no body.

Hi @IT Team , it will depend on your web framework. We use ASP.Net Core, so we just read the content body as

using var reader = new StreamReader(req.Body, Encoding.UTF8);
var content = await reader.ReadToEndAsync();

and then check to see if the content is empty or can be parsed as JSON. The transfer encoding is handled automatically.

Likewise if you are using node.js with express, you would use the `express.json()` middleware to transparently handle reading the content body as JSON data, and again the transfer-encoding is handled automatically.

I think that I am having the same issue as @IT Team

I had a webhook setup in Azure that has been running fine since 2021, accepting event notifications from RingCentral using the /telephony/sessions filter. As of yesterday, it appears that the requests are coming through with no bodies or I don't know how to access the bodies properly in their new form. The webhook is an Azure function app written in Node.js. I don't actually know javascript, I just knocked up the original webhook from code samples and a bunch of trial and error, so I don't really have the know how to take @Andrew Gaskill 's solution and potentially make it work for my situation. Does anybody have any suggestions for how I should alter my code to be able to access the data that contains the actual telephony sessions?

module.exports = async function (context, req) {
    context.log('JavaScript HTTP trigger function processed a request.');

    if (req.query.name) {
        context.res = {
            // status: 200, /* Defaults to 200 */
            body: "Hello " + (req.query.name)
        };
        context.bindings.outputQueueItem = req.body;
        context.log('200');
        context.log(req);
        context.done();
    }
    if (req.body  && req.body.uuid ) {
        context.res = {
            // status: 200, /* Defaults to 200 */
            body: "Hello " + (req.uuid )
        };
        context.bindings.outputQueueItem = req.body;
        context.log('200');
        context.log(req);
        context.done();
    }
    if ( req.url && req.headers['validation-token'] ) {
        context.res = {
            // status: 200, /* Defaults to 200 */
            headers: {"Validation-Token": req.headers['validation-token']},
            body: "Hello " + (req.url)
           
        };
        context.bindings.outputQueueItem = req.body;
        context.log('200');
        context.log(req);
        context.done();
    }
    if ( req.url && req.headers['verification-token'] ) {
        context.res = {
            // status: 200, /* Defaults to 200 */
            headers: {"Validation-Token": req.headers['verification-token']},
            body: "Hello " + (req.url)
           
        };
        context.bindings.outputQueueItem = req.body;
        context.log('200');
        context.log(req);
        context.done();
    }
    if ( req.url ) {
        context.res = {
            // status: 200, /* Defaults to 200 */
            body: "Hello " 
           
        };
        context.bindings.outputQueueItem = req.body;
        context.log('200');
        context.log(req);
        context.done();
    }
    else {
        context.res = {
            status: 400,
            body: "Please pass a name on the query string or in the request body"
        };
        context.log('400');
        context.log(req);
        context.done();
    }
};

Thanks for looking into that for me. I think that looks like a rabbit hole too far for an amateur like me to dig into. I'm going to try setting up my subscriptions with PubNub via the RingCentral SDK and hopefully that will be robust enough for my needs.

Switching to Nginx seems to fix things for us which I looked into after reading this:

https://www.jeffgeerling.com/blog/2017/apache-fastcgi-proxyfcgi-and-empty-post-bodies-chunked-transfer

Running PHP on Apache, we sometimes would get a request with the content-length header -- but not the transfer-encoding header -- and we could easily get the contents of the request

{
  "content-length": "1081"
  "user-agent": "RingCentral-WebHook",
  "accept-encoding": "UTF-8",
  "accept": "application/json",
  "content-type": "application/json; charset=UTF-8",
  "verification-token": "xxx",
  "connection": "close
}

And sometimes get a request with the transfer-encoding header -- but not the content-length header -- and we were unable to get the contents of the request

{
  "transfer-encoding": "chunked",
  "user-agent": "RingCentral-WebHook",
  "accept-encoding": "UTF-8",
  "accept": "application/json",
  "content-type": "application/json; charset=UTF-8",
  "verification-token": "xxx",
  "connection": "close"
}

When testing on Nginx, the headers now include both the transfer-encoding and content-length header and we are able to get the body of the request as we did before (shrugs)

{
  "transfer-encoding": "chunked",
  "user-agent": "RingCentral-WebHook",
  "accept-encoding": "UTF-8",
  "accept": "application/json",
  "content-type": "application/json; charset=UTF-8",
  "verification-token": "xxx",
  "connection": "close",
  "content-length": "1081"
}

For some users, the issue might be caused by apache server. So no matter how you change your code, it simply doesn't work. You may need to update your apache server or change to nginx.

Ref: https://bz.apache.org/bugzilla/show_bug.cgi?id=63366

Hi Developers,

I just want to let you know that we just rolled back the previous HTTP library which supports the content-length header. This roll-back is necessary and will be just a temporary solution for a certain period of time (about 30 days) so that developers would have enough time to prepare and update their web server that supports the transfer-encoding: chunked header.

For those who already successfully updated to newer web server that supports the transfer-encoding: chunked header should keep using that new web server since it should support both headers.

We will send out an official announcement with better information related to the change and timeline.

We apologize for any inconvenience this may have caused you due to this unintentional break.

Hi Iain et all,

I just had a discussion with the engineering team and the decision is that we will change the config to keep using the content-length header. So we will not roll back to use transfer-encoding: chunked anymore.

However, there is one remaining thing for us to decide is to add the body with some default content to the validation token post, the very first post that we send a validation-token header and the client app has to echo back the validation-token in the response header, in order to successfully create the webhook subscription. The reason we would add some content to the body is to keep the content-length header and set it to the length of the body content instead of content-length: 0.

I will announce when the final decision is made but you can assure to keep your code as it is right now.