Docker Compose桥接网络下容器TCP连接失败问题求助

Let's break down why your connection attempts are failing and walk through practical fixes step by step:

1. The Most Common Culprit: Service Listening on localhost (127.0.0.1) #

If your autoserver pymodbus service is configured to listen only on 127.0.0.1 (the container's loopback interface), it will only accept connections from inside its own container—even if other containers are on the same bridge network.

To fix this, you need to configure your server to listen on 0.0.0.0 (all available network interfaces) so it can accept connections from other containers in the network.

For example, in your autoserver's Python code:

from pymodbus.server.sync import StartTcpServer

# Replace 127.0.0.1 with 0.0.0.0 to accept external connections
StartTcpServer(your_context, address=("0.0.0.0", 5020))

2. Verify Network Connectivity Between Containers #

First, confirm both containers are attached to the same monitor-net network:

• Run docker inspect autoserver | grep Networks and docker inspect clientperf | grep Networks—both should list monitor-net as their active network.

Next, test DNS resolution and basic connectivity from the clientperf container:

# Exec into the clientperf container
docker exec -it clientperf bash

# Ping the autoserver by name (Docker's internal DNS should resolve this to its IP)
ping autoserver

# Test if port 5020 is open using netcat
nc -zv autoserver 5020

If ping works but nc fails, the issue is definitely with the autoserver's service not listening correctly (see step 1). If ping fails, double-check your Docker Compose config to ensure both services are explicitly attached to monitor-net.

3. Fix the Race Condition with depends_on #

The depends_on directive only ensures the autoserver container starts before clientperf—it doesn't wait for the pymodbus service inside autoserver to be fully ready. This means your client script might be trying to connect before the server finishes initializing.

Add a wait-for-ready logic to your clientperf service:

• Option 1: Use a tool like wait-for-it (install it in your clientperf Dockerfile):

  # In clientperf's Dockerfile
  RUN apt-get update && apt-get install -y wait-for-it
  

  Then update your Docker Compose command for clientperf:

  clientperf:
    # ... other config ...
    command: wait-for-it autoserver:5020 -- python your_script.py
  

• Option 2: Add a retry loop directly in your Python client code:

  from pymodbus.client.sync import ModbusTcpClient
  import time
  
  host = 'autoserver'
  client = ModbusTcpClient(host, port=5020)
  
  # Retry connection 5 times with 2-second delays
  connected = False
  for _ in range(5):
      if client.connect():
          connected = True
          break
      time.sleep(2)
  
  if not connected:
      raise Exception("Failed to connect to autoserver after retries")
  

4. Double-Check Port Listening Status #

Ensure your autoserver is actually listening on port 5020:

• Exec into the autoserver container and run:

  netstat -tulpn | grep 5020
  

  You should see output showing 0.0.0.0:5020 (not 127.0.0.1:5020). If nothing appears, your server isn't starting correctly—check the container logs with docker logs autoserver for startup errors.

Final Notes #

• You don't need ports mapping in your Compose config for inter-container communication—expose is sufficient (it declares the port is in use, and containers on the same network can access it directly).
• Using the service name autoserver is the correct approach (Docker's internal DNS resolves service names to container IPs in bridge networks)—avoid hardcoding IPs since they can change on container restart.

内容的提问来源于stack exchange，提问作者19mike95