Nginx-rtmp-module: already publishing when network broken.

[WayneShao]

When I use OBS to pushing the stream and break the network,Reuse the original address push stream after a few seconds,will not be able to push the stream successfully.

the log.

2017/11/20 17:01:25 [info] 8899#0: *7 client connected '192.168.159.1'
2017/11/20 17:01:25 [info] 8899#0: *7 connect: app='live' args='' flashver='' swf_url='' tc_url='rtmp://192.168.159.129/live' page_url='' acodecs=0 vcodecs=0 object_encoding=0, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:01:25 [info] 8899#0: *7 createStream, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:01:25 [info] 8899#0: *7 publish: name='default12312' args='' type=live silent=0, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:01:25 [error] 8899#0: *7 live: already publishing, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:02:52 [info] 8899#0: *9 client closed connection while waiting for request, client: 192.168.159.1, server: 0.0.0.0:5000
2017/11/20 17:05:59 [info] 8899#0: *7 recv() failed (104: Connection reset by peer), client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:05:59 [info] 8899#0: *7 disconnect, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:05:59 [info] 8899#0: *7 deleteStream, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:06:07 [info] 8899#0: *11 client connected '192.168.159.1'
2017/11/20 17:06:07 [info] 8899#0: *11 connect: app='live' args='' flashver='' swf_url='' tc_url='rtmp://192.168.159.129/live' page_url='' acodecs=0 vcodecs=0 object_encoding=0, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:06:07 [info] 8899#0: *11 createStream, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:06:07 [info] 8899#0: *11 publish: name='default12312' args='' type=live silent=0, client: 192.168.159.1, server: 0.0.0.0:1935
2017/11/20 17:06:07 [error] 8899#0: *11 live: already publishing, client: 192.168.159.1, server: 0.0.0.0:1935

my config file:

rtmp_auto_push on; # 切换自动推送(多 worker 直播流)模式
rtmp_auto_push_reconnect 1s; # 当 worker 被干掉时设置自动推送连接超时时间。默认为 100 毫秒。
rtmp { # 保存所有 RTMP 配置的块。
server { # 声明一个 RTMP 实例。
listen 1935; # 监听的端口号
# chunk_size 4096; # 流整合的最大的块大小。默认值为 4096。
ping 15s;
ping_timeout 10s;
application vod { # 创建一个 RTMP 应用。
play /usr/local/nginx/html/live; # 点播文件路径
}
application live { # 创建一个 RTMP 应用。
live on; # 是否直播
hls on; # 是否开启hls
hls_path /usr/local/nginx/html/live; # 设置 HLS 播放列表和分段目录。
hls_fragment 10s; # 设置 HLS 分段长度。
max_connections 1024; # 最大连接数
hls_playlist_length 30s; # HLS 播放列表长度
hls_sync 100ms; # HLS 时间戳同步阈值
meta copy; # 是否发送元数据到客户端
recorder manual { # 创建一个录制应用
record all manual; # 设置录制模式
record_suffix %Y-%m-%d-%H_%M_%S.flv; # 设置录制文件名
record_max_size 6200000K; # 设置录制文件的最大值
record_path /usr/local/nginx/html/Rec; # 指定录制的 flv 文件存放目录
}
}
}

[supercooker]

hi，I met the same problem.the camera connected to nginx server,and it worked well. if i shut down the power and reboot the camera ,it can not reconnect to the server. the stat.xml shows that there is still a connection with same ip .it seems that powering down the camera dosen't make nginx server disconnect the link.how do you solve the problem?thanks

[winshining]

It seems that the connection didn't close normally, so the server couldn't change the status of publisher. Directive drop_idle_publisher maybe help.

[supercooker]

yeah，I solved the problem last night , Directive drop_idle_publisher is useful,thanks for your answer!