Random deleteStream, client: 127.0.0.1, server: 0.0.0.0:1935 and signal 17 (SIGCHLD) received

[OnkelTem]

Hi!

I experience random stream resets.
This is my nginx.conf:

user www-data;
worker_processes 1;

pid /var/run/nginx.pid;
error_log /var/log/nginx/debug.log info;

events {
  worker_connections 1024;
}

rtmp {
  server {
    listen 1935;
    ping 20s;
    notify_method get;
    #chunk_size 4000;

    application full {
      # enable live streaming
      live on;
      allow publish 127.0.0.1;
      deny publish all;
      exec_pull ffmpeg -i rtsp://1.2.3.4:554 -c copy -f flv rtmp://localhost:1935/full/go;
    }
  }
}

When opening a page with player I get this in logs:

2014/07/17 14:18:19 [info] 2343#0: *1 client connected '22.33.44.55'
2014/07/17 14:18:19 [info] 2343#0: *1 connect: app='full' args='' flashver='LNX 14,0,0,125' swf_url='http://neganov.ru/js/jwplayer.flash.swf' tc_url='rtmp://neganov.ru:1935/full/' page_url='http://neganov.ru/home.html' acodecs=3575 vcodecs=252 object_encoding=0, client: 22.33.44.55, server: 0.0.0.0:1935
2014/07/17 14:18:19 [info] 2343#0: *1 createStream, client: 22.33.44.55, server: 0.0.0.0:1935
2014/07/17 14:18:19 [info] 2343#0: *1 play: name='go' args='' start=0 duration=0 reset=0 silent=0, client: 62.84.112.224, server: 0.0.0.0:1935
2014/07/17 14:18:19 [info] 2343#0: exec: starting managed child 'ffmpeg'
2014/07/17 14:18:20 [info] 2343#0: *2 client connected '127.0.0.1'
2014/07/17 14:18:20 [info] 2343#0: *2 connect: app='full' args='' flashver='' swf_url='' tc_url='rtmp://localhost:1935/full' page_url='' acodecs=0 vcodecs=0 object_encoding=0, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:18:20 [info] 2343#0: *2 createStream, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:18:20 [info] 2343#0: *2 publish: name='go' args='' type=live silent=0, client: 127.0.0.1, server: 0.0.0.0:1935

and video strarts playing OK. But then for no visible reason publishing stream just deleted, ffmpeg process is killed and restarted:

2014/07/17 14:18:45 [info] 2343#0: *2 deleteStream, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:18:45 [info] 2343#0: *2 disconnect, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:18:45 [info] 2343#0: *2 deleteStream, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:18:45 [notice] 2343#0: signal 17 (SIGCHLD) received
2014/07/17 14:18:45 [notice] 2343#0: unknown process 2347 exited with code 1
2014/07/17 14:18:45 [info] 2343#0: exec: child 2347 exited; ignoring
2014/07/17 14:18:45 [info] 2343#0: exec: terminating child 2347
2014/07/17 14:18:50 [info] 2343#0: exec: starting managed child 'ffmpeg'
2014/07/17 14:18:51 [info] 2343#0: *3 client connected '127.0.0.1'
2014/07/17 14:18:51 [info] 2343#0: *3 connect: app='full' args='' flashver='' swf_url='' tc_url='rtmp://localhost:1935/full' page_url='' acodecs=0 vcodecs=0 object_encoding=0, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:18:51 [info] 2343#0: *3 createStream, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:18:51 [info] 2343#0: *3 publish: name='go' args='' type=live silent=0, client: 127.0.0.1, server: 0.0.0.0:1935

and again:

2014/07/17 14:22:35 [info] 2343#0: *3 deleteStream, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:22:35 [info] 2343#0: *3 disconnect, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:22:35 [info] 2343#0: *3 deleteStream, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:22:35 [notice] 2343#0: signal 17 (SIGCHLD) received
2014/07/17 14:22:35 [notice] 2343#0: unknown process 2348 exited with code 1
2014/07/17 14:22:35 [info] 2343#0: exec: child 2348 exited; ignoring
2014/07/17 14:22:35 [info] 2343#0: exec: terminating child 2348
2014/07/17 14:22:40 [info] 2343#0: exec: starting managed child 'ffmpeg'
2014/07/17 14:22:41 [info] 2343#0: *4 client connected '127.0.0.1'
2014/07/17 14:22:41 [info] 2343#0: *4 connect: app='full' args='' flashver='' swf_url='' tc_url='rtmp://localhost:1935/full' page_url='' acodecs=0 vcodecs=0 object_encoding=0, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:22:41 [info] 2343#0: *4 createStream, client: 127.0.0.1, server: 0.0.0.0:1935
2014/07/17 14:22:41 [info] 2343#0: *4 publish: name='go' args='' type=live silent=0, client: 127.0.0.1, server: 0.0.0.0:1935

and again...
Any suggestions about these suicides?

[puneetdabulya]

I am also facing this problem. Did you find any solution? ffmpeg is getting killed without any reason in nginx logs.

[tgozed]

I used to have this issue. In the end I moved servers to a different hosting provider and that resolved the issue. I can only assume it was something to do with the firewall of my previous VPS provider.

[guoyi8]

I am also facing this problem...
Can someone help me with it ,please?

[misiek08]

I'm not sure, but I had similar problem and it was ffmpeg dying itself and causing nginx to reset it. So it was fully on ffmpeg side, randomly crashing after some long time.

[syrkandonut]

1. Check the ffmpeg command, or remove it altogether for a while
2. Make sure that the RTMP source is sending the data correctly

Even though it's 2025, I'm still here. Indeed, this error may be related to ffmpeg and its commands, for example, when lib is missing for ffmpeg. But also, we might have missed that the RTMP source is sending incorrect data. For example, in my case, the IP camera sent an RTMP stream with h265 compression, although this is not supported by the standard RTMP protocol. I replaced the compression in the camera with H264 and everything worked.