本地Docker容器突然出现Python socket.gaierror：无法解析容器地址问题排查求助

Let’s break down why your socket.gethostbyname("container_name") is throwing socket.gaierror locally (but working in GitLab CI) and how to diagnose and fix it:

First, Verify Basic DNS Resolution #

Start with simple manual checks to isolate whether the issue is with Docker’s DNS or your Python code:

• Run nslookup container_name or ping container_name in your local terminal. If this fails to return the container’s IP, the problem lies with Docker’s built-in DNS service, not your test script.
• If the ping works but Python still throws an error, skip ahead to the Python-specific debugging steps.

Check Docker Network & DNS Configuration #

Docker Compose creates a dedicated network for your services by default, which handles container name resolution. Here’s what to inspect:

• Confirm the container is in the expected network: Run docker inspect container_name | grep Networks to see which network your container uses. Then run docker network inspect [network-name] and look for your container’s name under the Containers section—it should have a DNS entry mapped to its IP address.
• Restart Docker: Network cache glitches are a common culprit. Restart Docker Desktop (or the Docker daemon on Linux) and retry your test.
• Validate Docker daemon DNS settings: For Linux, edit /etc/docker/daemon.json and ensure it includes valid DNS servers (e.g., {"dns": ["8.8.8.8", "8.8.4.4"]}). On Docker Desktop, go to Settings > Docker Engine to verify the same. After making changes, restart the daemon.

Clear Local DNS Cache & Hosts File #

Stale local DNS data or incorrect hosts entries can block container name resolution:

• Flush DNS cache:
  • Mac: sudo dscacheutil -flushcache; sudo killall -HUP mDNSResponder
  • Windows: ipconfig /flushdns
  • Linux: sudo systemd-resolve --flush-caches
• Inspect hosts file: Check /etc/hosts (Linux/Mac) or C:\Windows\System32\drivers\etc\hosts (Windows) for any stale or incorrect entries for container_name—delete them if present.

Debug Python’s DNS Resolution #

Python might be using unexpected DNS settings that differ from your terminal’s configuration:

• Add debug code to your test to see how Python is attempting to resolve the name:

  import socket
  try:
      print("Resolving container_name...")
      addr = socket.gethostbyname("container_name")
      print(f"Resolved to: {addr}")
  except socket.gaierror as e:
      print(f"Error details: {e}")
      # Check environment variables that might affect DNS
      import os
      print(f"System DNS (from env): {os.environ.get('DNS_SERVER', 'Not set')}")
  

• Test with the container’s direct IP: Run docker inspect container_name | grep IPAddress to get the container’s IP, then replace container_name with this IP in your test. If this works, the issue is definitely related to name resolution.

Quick Workarounds If You Need Immediate Results #

If you can’t pinpoint the root cause right away:

• Assign a static IP to the container: Update your docker-compose.yml to set a fixed IP address for the container:

  version: "2.3"
  networks:
    default:
      ipam:
        config:
          - subnet: 172.20.0.0/16
  services:
    container_name:
      image: container_image:latest
      build:
        context: ./
      container_name: container_name
      ports:
        - "10021:10021"
        - "10022:10022"
        - "10023:10023"
      restart: unless-stopped
      networks:
        default:
          ipv4_address: 172.20.0.10
  

  Then use 172.20.0.10 directly in your test, or add 172.20.0.10 container_name to your local hosts file to restore name-based access.

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