容器内的docker pull图像因超时而失败

Question

I am running a docker-compose that has 3 services. My main service depends on an image service which depends on a dind-service

  service-1:
    image: 'someimage'
    privileged: true
    ports:
      - '8911:8911'
    environment:
      - docker-compose=true
    volumes:
      - scratch:/tmp/scratch
      - bundles:/tmp/bundles:ro
    command: http-compile
    depends_on:
      - img

  img-service:
    image: 'img'
    privileged: true
    ports:
      - '8909:8909'
    environment:
      - DOCKER_HOST=tcp://dind-img:2375
    volumes:
      - bundles:/tmp/bundles
      - ./deploy/path/to/config.json:/etc/config.json
    command: runc-setup
    depends_on:
      - dind-img

  dind-service:
    image: docker:18-dind
    privileged: true
    volumes:
      - execimages:/var/lib/docker
      - tmp:/tmp

All these services images are built locally but one of the services fetches other images (like language runtimes, for example Go) from the net and builds them locally inside the container.

But while fetching the image I get timeout errors and because of that my service which depends on the other two services fails.

I was doing this from my workplace and I tried to see if we had a proxy configured but running the command env | grep proxy env | grep proxy gave me nothing.

So it is likely not a network problem, maybe something to do with my docker-setup.

folks at my workplace who develop this project on Mac have never faced this issue with docker pull. I am on Ubuntu 18.04

I ran a shell script from the docker repol to check the config to check whats missing and these were all the missing results. Are any of these causing the problem?

CONFIG_NF_NAT_IPV4: missing
CONFIG_NF_NAT_NEEDED: missing
CONFIG_MEMCG_SWAP_ENABLED: missing
CONFIG_IOSCHED_CFQ: missing
CONFIG_CFQ_GROUP_IOSCHED: missing
CONFIG_RT_GROUP_SCHED: missing
CONFIG_INET_XFRM_MODE_TRANSPORT: missing
zfs command: missing
zpool command: missing

Answer 1

Check your logs to see if you see anything like this:

[0683] Handler for POST /images/create returned error: Get https://registry-1.docker.io....: dial tcp: lookup registry-1.docker.io on 8.8.8.8:53: read udp 192.168.0.163:32769->8.8.8.8:53: read: connection refused (or timeout)

This suggests that the DNS query is being run from an ephemeral port on your ubuntu and DNS servers either timeout or reject the request.

Instead add your local DNS cache/server as the primary on resolve.conf eg cat /etc/resolv.conf

nameserver 192.168.0.1
nameserver 8.8.8.8
....

Then your docker pull should work successfully.