[from fork parent #22] Allow AES67 Subscriptions

[rgon]

See @Mo-way's fork: chris-ritsen#22 (comment)
can be found at: https://github.com/Mo-way/network-audio-controller/tree/feat/create-aes67-multicast-channels
It purportedly does not work? -> test locally:

It created a Multicast Flow, but not an AES67 Multicast Flow.
If 1234 is used.
It created an AES67 Flow if 12 is used.
And the command overwrites the flow form before.

Aparently it doesn't handle >2 ch properly

[Mo-way]

I tested with 2 channel AVIO only. If you have anything with more channels, please do test!

[jayxcfgj]

@Mo-way I have access to an x32 with an dante expansion card which is an Brooklyn II chip. This device has 32 channels and its capable of setting it in the AES67-mode. So I can do tests for you to figure it out. But you need to tell me what i should do exactly. If i understand it right we execute the actions in the Dante controller that we want to implement and capture the network traffic via wireshark and extract some data of the traffic, I'm right?
I really want to see this AES67 routing thing working, because this is the last puzzle piece we need to control the AOIP under linux completly.

[Mo-way]

@jayxcfgj Yes, exactly. Please do the wireshark captures!

I described it here: chris-ritsen#22 (comment)

If it doesn't work right away, just send a full network capture of you doing the actions + timestamp when you klicked which button. I can try filtering out the noise myself.

[rgon]

@jayxcfgj yes please! I was about to purchase a Dante device just for this, even despite the cost, but work got in the way.

Steps to submit a network capture

Prerequisites if you'll use a linux device to run wireshark: just install it (for Ubuntu, run sudo apt update && sudo apt install wireshark && sudo usermod -aG wireshark "$USER" && sudo wireshark

1. Build the Dante Network:

to an unmanaged/dumb switch:
  A: computer with wireshark (sniffing traffic)
  B: x32

  C: client Dante/AES67 or AES67-compatible device (so no brooklyn I chips!)
  D: windows computer with Dante controller

To reduce the device count, the C controller computer could be a virtualbox VM (tip: use the free 'MS Edge Virtualbox VM', DL from the Internet Archive if it's down), and the B client device I suppose could be any of the dante virtual soundcards or the ravenna AES67 driver/other. The D sniffing computer could be the VM host. So the minimal setup could be:

  A: Linux PC with
    + wireshark
    + Virtualbox with Windows VM
        + Dante Controller
        + Dante virtual soundcard
  B: x32

2. Open Dante Controller

Perform any prerequisite config (enabling aes67, removing connected channels) now, we don't need that captured.

3. Open wireshark on the linux computer

maybe it also works on a windows host?

4. Set it to listen on the ethernet device that's connected to the Dante network

Probably enp_s____ or enx____________`

5. Apply filters to capture only Dante-relevant traffic

@Mo-way helpfully specified them in chris-ritsen#22 (comment)

6. Perform the explicit tasks we need:

'connecting AES67 x32 ch 8 to client ch 2' and record the timestamp (eg 01:39) and perform as many variations as reasonable, logging their timestamps. It'd be easiest if there were, for example, 5 changes in 5 different minute timestamps.

7. Finish and save

Once finished, press the 'stop' button to stop capturing noise and save the capture to file.

8. Upload the .pcapng file here.

---

The general steps 1-5 are always the same for any Dante package capture.

[jayxcfgj]

okay so attached there are a bunch of data-captures and a few screenshots so you get an idea what i did, but i think with the filenames you will get it. If you have one dataset that doesnt fit to the rest, don't get headache and skip this dataset and take the next one. Maybe i forgot one or two times to click the radio button to switch from dante multicast flow to aes67 multicast flow. i must did this every time again if i created a new flow multicast. I hope you find this useful.

captures.tar.gz

[Mo-way]

@jayxcfgj Thanks, that looks promising!

In order to see the commands sent, I did this filter:

ip.dst == 169.254.188.183

I assume:

• 169.254.188.183 belongs to the x32?
• 169.254.136.111: the PC running Dante Controller

[jayxcfgj]

@Mo-way Yes this is correct.

[Mo-way]

I had a look at my old implementation and some things don't add up.
There seem to be magic fields that are constant in your run, but they differ from the fields in another capture I acquired.

Wrote unit test with data copied from wireshark: https://github.com/Mo-way/network-audio-controller/blob/feat/create-aes67-multicast-channels/tests/test_aes67.py

@jayxcfgj I like the way you made the captures, were easy to work with. If there's time, could you create more captures where you activate:

• last channel (32)
• two last channels (31,32)
• second to last channel (31)
• optional: ch30 alone as well
• third channel (3)
• fourth channel (4)

Would be great!
Chef's kiss if you filter with !upd.port == 5004 to exclude the RTP traffic.

[jayxcfgj]

@Mo-way okay so here comes the captures Part II. I have worked through all your points, but I haven't had any success with your exclude rtp filter. If i paste !upd.port == 5004 in the filter row in the main window of wireshark i get an error. To check if the relevant stuff is captured successfully, I usually paste the phrase udp.port == 4440 || udp.port == 8700 in the filter row.
Have I forgotten or done something wrong?

captures II.tar.gz

[Mo-way]

@jayxcfgj Thanks a lot.
Sorry, for the typo, My bad. Of course it's udp, not uPd .

Your approach is totally fine as well, I was just lazy, so I don't have to run udp.port == 4440 || udp.port == 8700

[Mo-way]

@jayxcfgj Got unit tests passing for creating any amount of streams: https://github.com/Mo-way/network-audio-controller/tree/feat/create-aes67-multicast-channels

Currently on the road, will test with an avio soon. Usage, if you'd like to try:

git clone https://github.com/Mo-way/network-audio-controller.git
git checkout feat/create-aes67-multicast-channels
poetry run netaudio config --aes67-activate-multicast "1 2 3 4"
poetry run netaudio config --aes67-activate-multicast "8 10"

[jayxcfgj]

@Mo-way Okay today I tested your new implementation and didn't get working. If I type the command i get a new "Prompt" " (Pdb) " (as you can see in my screenshot). From there i don't know what i should type in. The amount of channels for the muslticast flow? Why? If i try this nothing happens no matter what i typeing in. I think something is wrong or I'm dumb and don't get it.

By the way: While reading through the help section, I noticed that a function is still missing to apply the command for activating/deactivating aes67 mode. And this is a reboot of the dante-device. So I did a quick wireshark capture for this function if you want to implement that, too.

reboot-dante-device.pcapng.gz

[Mo-way]

@jayxcfgj Sorry about the (PDB), I forgot a breakpoint() in the code. You can continue running by pressing c. I also pushed a fixed version.

(It's about time we get a CI for this, ruff format, ruff check etc. then those stupid mistakes are caught)

Thanks, for the hint. Enabling/disabling is already merged by chris. My branch needs to rebase on his changes, but that's a pain, so I skipped doing so for now...

[jayxcfgj]

@Mo-way okay, so i gave it another try today with your latest code and I got some interesting reuslts:
At first, it worked fine! As you can see in the first screenshot, i was able to create an RTP stream with multiple channels without any problems.

However, when I tried to add more channels, I expected, that it will create a second rtp stream.
But what happend is that it creates a new one and it replaced the first one immediately. (See the second screenshot.)

And here is an other thing. There is a limitation on the number of channels that can be contained in an RTP stream. If you take a look at my capture screenshots, you will see a hint in the window where I choose which channels to transive. In my case, the limitation is 8 channels per stream. I think this is specific to the device, but I'm not sure...
However if I try to create more than the acceptable amount of channels in one command, it fails and all the existing streams are cancelled. I don't know how to prevent the user from doing this, because for this you need to nknow the limitation... In the dante controller it splits the channels into multiple rtp streams automatically....

We also need a command to delete existing rtp-streams. To do this, the user needs to specify which stream they want to delete. For this i think we also need a listing command to tell the user which channels and which streams are active.

By the way, I also tested the AES67 mode enable/disable function, and it worked very well!