Debugging live streaming ingest protocol

Question

Hi everyone,
I posted a separate question last week about live streaming protocols. I had been searching for documentation on fmp4 live streaming ingestion and found nothing, but afterward I found the Smooth Streaming protocol specification (MS-SSTR-- the forum won't
let me link to it directly), which seems to apply to my use case. Now I've spent a few days trying to make it work and so far haven't been able to do so.
I'm not looking for a canned solution; I'm looking for resources to help me debug the problem.
According to the specification, I should be able to make a long-lived, chunked http POST request to my ingest endpoint, sending the fragmented mp4 data as the request body. I've tried to implement this without success-- I open the request stream and
write 1000 byte chunks at a time, but the connection is reset after only one or two chunks. I can't tell if the problem is the data or the chunk size or something else.
And if I mimic Expression Encoder 4, which seems to diverge from the spec and not send its data chunked at all, the connection isn't reset no matter how much data I send, but the video never makes it to the preview stream on the Azure portal.
So the question is, how can I debug this? Is there any error log or diagnostic output I can access? Each time I try something new, it doesn't work, but I have no way of knowing whether I am getting closer to or further from a working solution. I'm a bit
lost here, so help would be greatly appreciated.

Answer 1

Hi Jason, and thanks for getting back to me.
The IP security thing is a good thought, but I think I have that part taken care of already. I saw the option to restrict ingest to the IP I used to configure the channel from in the Portal, and I made sure to set it to Off. Also, I can stream successfully
from Expression, which would have the same source IP.
As for why I want to use Smooth Streaming, basically there are two reasons. First, it seems like the simpler protocol conceptually. I'm already generating fragmented mp4 files (though I think they might be packaged wrong; I can't really tell which part is
causing problems), and it seems simple enough that I could implement it without turning to a third-party library (the part that's feeding the stream is written in C#, though I'd be willing to rewrite in C++ if necessary, and there's no RTMP client
that seems trustworthy-- FluorineFX, for instance, seems to have ceased development years ago now).
Second, while reading the RTMP spec, I realized that even if I had such a library it isn't clear how I would need to use it with Azure. It allows creating objects or invoking methods on the server, but I couldn't find any guide to which objects and methods
might be supported. And the spec also doesn't define specifically what video format, codec options, or packaging method I need to use; essentially I would seem to have all the same problems I'm having with Smooth, only with the added difficulty of needing
a third-party library to implement the spec properly.
I guess the main problem is that I don't really know how close or far I am from success. Like I said, I think I am doing something wrong in how the video is encoded, but for all I know I'm being dropped because I'm feeding the data too quickly
or slowly. (I can't find any documentation on how much data Azure can buffer or how it handles underruns when the video to be streamed live is generated on the fly and not tied to strictly realtime output). If there's more documentation on specifically how
Azure uses or expects data from these protocols, it'd be very helpful.