如何在非Marathon的Mesos框架下启动带端口映射的Docker容器？

Got it, let's break down your problem into two core parts: exposing custom ports for Docker containers via Mesos, and resolving bridge network startup issues—all without relying on Marathon. Here's a step-by-step solution tailored to your scenario:

1. Exposing Custom Ports in Mesos Docker Tasks #

Mesos handles port allocation as a dedicated resource, so you need to explicitly request ports in your task definition, then map them to your Docker container's internal ports. There are two common approaches: fixed ports (you specify the exact host port) or dynamic ports (Mesos assigns an available port from a range).

Fixed Port Example #

If you need a specific host port (e.g., 8080 mapped to container port 80), define your task like this:

{
  "id": "my-custom-docker-task",
  "resources": [
    {"name": "cpus", "type": "SCALAR", "scalar": {"value": 0.5}},
    {"name": "mem", "type": "SCALAR", "scalar": {"value": 256}},
    {"name": "ports", "type": "RANGES", "ranges": {"range": [{"begin": 8080, "end": 8080}]}}
  ],
  "container": {
    "type": "DOCKER",
    "docker": {
      "image": "your-image:tag",
      "network": "BRIDGE",
      "portMappings": [
        {
          "containerPort": 80,
          "hostPort": 8080,
          "protocol": "tcp",
          "name": "http"
        }
      ]
    }
  },
  "command": {"value": "your-container-start-command"}
}

Dynamic Port Example #

If you prefer Mesos to pick an available port from a range (e.g., 8000-9000), set hostPort: 0 and define a port range in resources:

{
  "id": "my-dynamic-port-task",
  "resources": [
    {"name": "cpus", "type": "SCALAR", "scalar": {"value": 0.5}},
    {"name": "mem", "type": "SCALAR", "scalar": {"value": 256}},
    {"name": "ports", "type": "RANGES", "ranges": {"range": [{"begin": 8000, "end": 9000}]}}
  ],
  "container": {
    "type": "DOCKER",
    "docker": {
      "image": "your-image:tag",
      "network": "BRIDGE",
      "portMappings": [
        {
          "containerPort": 80,
          "hostPort": 0,
          "protocol": "tcp",
          "name": "http"
        }
      ]
    }
  },
  "command": {"value": "your-container-start-command"}
}

To get the dynamically assigned port, query the Mesos API (e.g., GET /tasks) and check the task's resources.ports or status.containerStatus.networkInfos fields.

2. Resolving Bridge Network Startup Issues #

If you can't start containers in bridge mode, here are the key fixes to check:

• Verify Mesos Agent Configuration: Ensure your Mesos agent isn't forced to use host network. Check the agent's startup flags—remove any --docker-network=host flag, or explicitly set --docker-network=bridge to default to bridge mode.
• Check Docker Daemon Bridge Network: Run docker network inspect bridge on the agent host to confirm the default bridge network exists and has valid subnet/gateway settings. If it's corrupted, reset it with docker network prune followed by restarting the Docker daemon.
• Firewall & Port Conflicts: Make sure the host firewall (e.g., iptables, ufw) isn't blocking traffic between the host and bridge network. Also, if using fixed ports, confirm no other process is using that port on the agent host.
• Legacy Mesos Versions: For older Mesos versions (pre-1.0), you might need to pass bridge network via Docker parameters instead of the network field:

  "docker": {
    "parameters": [
      {"key": "network", "value": "bridge"},
      {"key": "publish", "value": "8080:80"}
    ]
  }
  

Final Notes #

• Always validate your task definition with the Mesos API before submitting to catch syntax errors.
• If you're using a custom Mesos framework, ensure it properly handles port resources and Docker network settings in its task scheduler logic.

内容的提问来源于stack exchange，提问作者Alisa Alisa