owt-server 服务开机自启

owt-server 服务开机自启

owt-server 服务需要进入镜像内部，执行startowt.sh 脚本进行启动。但如果服务器断电重启了，服务在业主地方。业主又是内网环境，启动服务就会很复杂。 这个时候 owt-server 开机自启就很重要了，官网并没有开机自启的方式。研究了一下可以实现，需要改一下 docker启动方式。

目前只用到4.3 和 5.1.0 这两个版本的。

owt-server 4.3 开机自启方式

4.3 版本没有提供 startowt.sh 脚本，需要自己添加启动方式的 shell 脚本

docker run -it  --restart=always   -d  --name owt-server -p 3004:3004 -p 3300:3300 -p 8092:8092 -p 60000-60050:60000-60050/udp \
    -v /home/zbh/owt-server/portal/portal.toml:/home/owt/portal/portal.toml \
    -v /home/zbh/owt-server/webrtc_agent/agent.toml:/home/owt/webrtc_agent/agent.toml \
    -v /home/zbh/owt-server/logs:/home/owt/logs \
    -v /home/zbh/owt-server/startowt.sh:/home/startowt.sh \
    yuhanghate/owt-server:4.3.1 /bin/bash /home/start.sh

这里的start.sh 脚本是我自己做镜像时挂上去的。如果需要改动也可以映射出来改。

• start.sh脚本内容：

  #!/usr/bin/env bash
  /home/owt/bin/init-all.sh && /home/owt/bin/start-all.sh
  sleep infinity

这里主要是加了 sleep infinity 命令。可以让主机重启以后，运行的docker 镜像重启服务器以后。可以一直不结束，如果不加这个命令。会导致docker镜像重启了，里面的服务没有重启。3004或者3300对应端口的网站没有启动

owt-server 5.1.0 开机自启方式

    docker run -it  --restart=always  --user root  -d  --name owt-server -p 3004:3004 -p 3300:3300 -p 8092:8092 -p 60000-60050:60000-60050/udp \
    -v /home/zbh/owt-server/portal/portal.toml:/home/owt/portal/portal.toml \
    -v /home/zbh/owt-server/webrtc_agent/agent.toml:/home/owt/webrtc_agent/agent.toml \
    -v /home/zbh/owt-server/logs:/home/owt/logs \
    -v /home/zbh/owt-server/startowt.sh:/home/startowt.sh \
    yuhanghate/owt-server:5.1.0 /bin/bash /home/startowt.sh

把 /home/zbh/owt-server 路径换成实体机上的路径名称。启动的话，需要配置portal.toml和agent.toml两个文件。
镜像是owt 用户，需要加 --user root 把当前镜像升级到root权限。不然执行 startowt.sh脚本的时候，会失败。

• portal.toml 文件配置

  [portal]
  keystorePath = "./cert/certificate.pfx"
  hostname = "" #default: ""
  ip_address = "192.168.100.210" #default: ""
  # Port that the socket.io server listens at.
  port = 8092 #default: 8080
  ssl = true #default: true
  force_tls_v12 = false #default: false, tls1 and tls1.1 will not be allowed if set to true
  #The network inferface the socket.io will be bound to in case portal.ip_address (the 'ip_address' item above) is not specified or equal to "".
  # networkInterface = "eth1"
  # How many seconds before Socket.IO client sending a new ping packet. This value might be overwrote by client side.
  # ping_interval = 25 #default: undefined
  # How many seconds without a pong packet to consider the Socket.IO connection closed. This value might be overwrote by client side.
  # ping_timeout = 60 #default: undefined
  # Life time for reconnection ticket. Unit: second.
  reconnection_ticket_lifetime = 600 #default: 600
  # Client will leave conference if it cannot reconnect to Socket.IO server after |reconnection_timeout| seconds.
  reconnection_timeout = 5 #default: 60
  #default: *, allow cross origin request share list. Add http(s)://web-app-server-ip-or-hostname:port to cors list in product deployment, example: ["http://web-app-server-ip-or-hostname:3001", "https://web-app-server-ip-or-hostname:3004"].
  cors = ["*"]
  ​
  ​
  [cluster]
  name = "owt-cluster"
  ​
  #The number of times to retry joining if the first try fails.
  join_retry = 60 #default: 60
  ​
  #The interval of reporting the work load
  report_load_interval = 1000 #default: 1000, unit: millisecond
  ​
  #The max CPU load under which this worker can taks new tasks.
  max_load = 0.85 #default: 0.85
  ​
  ​
  [capacity]
  #The ISP list this portal is able to handle
  #If the ISP list is set to be non-empty, only the creating token requests with preference.ips being matched with one element of this list will be scheduled to this portal.
  isps = [] #default: [], which means all ISPs.
  #The region list this portal prefers to handle
  #If the region list is set to be non-empty, the creating token requests with preference.region being matched with one element of this list will be priorly scheduled to this portal.
  regions = [] #default: [], which means all regions.
  ​
  ​
  [rabbit]
  host = "localhost" #default: "localhost"
  port = 5672 #default: 5672
  ​
  [mongo]
  dataBaseURL = "localhost/owtdb" #default: "localhost/owtdb"

ip_address 属性改成自己主机的ip地址

reconnection_timeout ：可以修改owt服务的websocket超时时间。防止杀进程等操作时2个画面时间过长

• agent.toml 配置文件

  [agent]
  #Max processes that agent can run
  maxProcesses = 13 #default: 13
  ​
  #Number of precesses that agent runs when it starts. 1 <= prerunProcesses <= maxProcesses.
  prerunProcesses = 2 #default: 2
  ​
  ​
  [cluster]
  name = "owt-cluster"
  ​
  #The number of times to retry joining if the first try fails.
  join_retry = 60 #default: 60
  ​
  #The interval of reporting the work load
  report_load_interval = 1000 #default: 1000, unit: millisecond
  ​
  #The max network load under which this worker can take new tasks.
  max_load = 0.85 #default: 0.85
  ​
  #The name of network-interface for reporting load.
  network_name = "lo"
  ​
  #The bandwidth of network-interface used for WebRTC peerconnections. 
  network_max_scale = 1000 #unit: Mbps
  ​
  ​
  [capacity]
  #The ISP list this agent is able to handle.
  #If the ISP list is set to be non-empty, only the creating token requests with preference.ips being matched with one element of this list will be scheduled to this agent.
  isps = [] #default: [], which means all ISPs.
  #The region list this agent prefers to handle
  #If the region list is set to be non-empty, the creating token requests with preference.region being matched with one element of this list will be priorly scheduled to this agent.
  regions = [] #default: [], which means all regions.
  ​
  ​
  [rabbit]
  host = "localhost" #default: "localhost"
  port = 5672 #default: 5672
  ​
  [internal]
  #The IP address used for internal-cluster media spreading. Will use the IP got from the 'network_interface' item if 'ip_address' is not specified or equal to "".
  ip_address = "" #default: ""
  ​
  #The network interface used for internal-cluster media spreading. The first enumerated network interface in the system will be adopted if this item is not specified.
  # network_interface = "eth0" # default: undefined
  ​
  # The internal listening port range, only works for TCP now
  maxport = 0 #default: 0
  minport = 0 #default: 0
  ​
  #########################################################################################
  [webrtc]
  #The network inferface all peer-connections will be established through. All network interfaces in the system will be adopted if this item is not specified or specified with an empty array.
  ##Valid value is an array of objects that has two properties: name and replaced_ip_address. name is the name of network interface that will be used, replaced_ip_address is the IP address used for replacing the internal IP address in locally generated SDP's and ICE candidates during establishing the peer-connection(useful when behind NATs). replaced_ip_address is optional, if it is not specified, IP address from corresponding network interface will not be replaced.
  #Example of valid value: [{name = "eth1"}, {name = "eth2", replaced_ip_address = "192.0.2.2"}].
  network_interfaces = [{name="eth0",replaced_ip_address="192.168.100.210"}]  # default: []
  ​
  keystorePath = "./cert/certificate.pfx"
  #note, this won't work with all versions of libnice. With 0 all the available ports are used
  ​
  # The webrtc port range
  maxport = 0 #default: 0
  minport = 0 #default: 0
  ​
  #STUN server IP address and port to be used by the server.
  #if "" is used, the address is discovered locally
  stunport = 0 #default: 0
  stunserver = "" #default: ""
  ​
  #ThreadPool worker numbers for peer connection
  num_workers = 24 #default: 24

network_interfaces = [{name="eth0",replaced_ip_address="192.168.100.210"}]把里面的 replaced_ip_address 换成本机的ip地址或者公网ip地址。

name : 就用默认的 eth0 ， 换别的好像会有问题