search

sistason / rtmp_stream_delayer

Uses ffmpeg and nginx to delay (buffer) an incoming rtmp-stream for an amount of time.
GNU General Public License v2.0
42 stars 12 forks source link

Again many troubles - sorry for bothering #11

Open Trauni opened 3 years ago

Trauni commented 3 years ago

Hello Sir,

we again expierencing a lot of troubles. Following is happening and it sounds like an network error but we only have this issues when we use the delay script. Maybe you can help us:

We are streaming to our Nginx-Server (a big 64GB RAM, i9 9900k machine on Ubuntu). Delay is working fine We also stream with 2 additional observer to the server without stream on different applications (but the error occurs also without them). After more then 1 hour suddenly the upstream from OBS is not able to stream anymore to the server, the bitrate goes down to 0

On the server I see in the folder with the pid-file that all x-seconds a new 0Byte Flv file appears. The delayer run into problem (Input wrong): Invalid data found when processing input

The sender is not able to restart OBS because the NGINX-RTMP Key is blocked. I have to stop the delayer, restart nginx and then all things will work on but it happens also again after a couple of time.

I really dont understand what is going on, also asked the Server provider to send me network statistics.

Best MicrosoftTeams-image

sistason commented 3 years ago

it looks as after a while, the nginx produces only 0KB streams anymore, or cannot write anymore.

The delayer fails ofc, as 0KB files get rejected by ffmpeg.

I would expect some buffers to be full, disk full, permissions changed, etc, so that nginx could not write your stream to the folder anymore, leading to this behaviour you see.

I am completely unknowing of the underlay of the nignx-rtmp stuff, if it works fine after a restart, there seems to be an issue with that? I'd be interested if you find the problem, I can only think of the more obvious stuff like disk full, but which is probably the first thing you checked...

Trauni commented 3 years ago

I found some things:

First of all in these cases the provied of my server got DDOS attacks.

After clearing the network routes the nginx-rtmp has only the problem that the key is blocked for about 10 minutes. So we are not able to simply restart the stream with the same key and then the delayer wont work on if we use a new key.

The provider does now better DDOS prevention and the last 5 days all productions where good. Hopefully we are safe right now. Therefore I now know its nothing about ffmpeg and nginx, its only those attacks.