docker_container pull 'Get http://: http: no Host in request URL'

[davegallant]

ISSUE TYPE

• Bug Report

COMPONENT NAME

• docker_container

ANSIBLE VERSION

• 2.2.0.0

OS
Ubuntu 14.04.5 LTS

Summary:
This docker_container module was working up until now and have not changed any code. When I try to pull the image manually in a ssh session, it works.

- name: Kazoo Install
  docker_container:
    name: soa-kazoo
    docker_api_version: "{{ docker_api_version }}"
    image: blah-testing.artifactory.internal/soa-kazoo
    env:
      ZOOKEEPER_PORT: "{{ zookeeper_host }}"
    log_driver: syslog
    command: /bin/bash -c 'export HOME={{ working_dir }}; ./run_api.sh'
    state: started

Error:
fatal: [api]: FAILED! => {"changed": false, "failed": true, "msg": "Error pulling blah-testing.artifactory.internal/soa-kazoo - code: None message: Get http://: http: no Host in request URL"}

[ansibot]

@Davey-Dev, we need more information to resolve this issue. For that reason, we put this Issue into the 'needs_info' state.

Here are the required items we could not find in your description:

• issue type

When you have filled in the missing data, we will notify the module maintainer for further action.
click here for bot help

[clane-axial]

I'm seeing same problem on Ubuntu 16.04.1.

[kurron]

I am also experiencing the issue. Like, @Davey-Dev , the same script worked yesterday but fails today. I did have to switch to a different VM so the environment is not identical but very close. Here is the dump I get trying to pull down the RabbitMQ image from Docker Hub with -vvv enabled:

Using module file /usr/local/lib/python2.7/dist-packages/ansible/modules/core/cloud/docker/docker_image.py
<localhost> EXEC /bin/sh -c '( umask 77 && mkdir -p "` echo $HOME/.ansible/tmp/ansible-tmp-1480534399.17-47607955013186 `" && echo ansible-tmp-1480534399.17-47607955013186="` echo $HOME/.ansible/tmp/ansible-tmp-1480534399.17-47607955013186 `" ) && sleep 0'
<localhost> PUT /tmp/tmp5rYCT3 TO /home/vagrant/.ansible/tmp/ansible-tmp-1480534399.17-47607955013186/docker_image.py
<localhost> EXEC /bin/sh -c 'chmod u+x /home/vagrant/.ansible/tmp/ansible-tmp-1480534399.17-47607955013186/ /home/vagrant/.ansible/tmp/ansible-tmp-1480534399.17-47607955013186/docker_image.py && sleep 0'
<localhost> EXEC /bin/sh -c '/usr/bin/python /home/vagrant/.ansible/tmp/ansible-tmp-1480534399.17-47607955013186/docker_image.py; rm -rf "/home/vagrant/.ansible/tmp/ansible-tmp-1480534399.17-47607955013186/" > /dev/null 2>&1 && sleep 0'
failed: [localhost] (item={u'image': u'rabbitmq', u'tag': u'3.6.5-management'}) => {
    "failed": true, 
    "invocation": {
        "module_args": {
            "api_version": null, 
            "archive_path": null, 
            "buildargs": null, 
            "cacert_path": null, 
            "cert_path": null, 
            "container_limits": null, 
            "debug": false, 
            "docker_host": null, 
            "dockerfile": null, 
            "filter_logger": false, 
            "force": false, 
            "http_timeout": 240, 
            "key_path": null, 
            "load_path": null, 
            "name": "rabbitmq", 
            "nocache": "False", 
            "path": null, 
            "pull": true, 
            "push": false, 
            "repository": null, 
            "rm": true, 
            "ssl_version": null, 
            "state": "present", 
            "tag": "3.6.5-management", 
            "timeout": null, 
            "tls": null, 
            "tls_hostname": null, 
            "tls_verify": null, 
            "use_tls": "no"
        }, 
        "module_name": "docker_image"
    }, 
    "item": {
        "image": "rabbitmq", 
        "tag": "3.6.5-management"
    }, 
    "msg": "Error pulling rabbitmq - code: None message: Get http://: http: no Host in request URL"
}

I know the module relies on py-docker and this is what pip tells me:

Name: docker-py
Version: 1.10.6
Summary: Python client for Docker.
Home-page: https://github.com/docker/docker-py/
Author: Joffrey F
Author-email: [email protected]
License: UNKNOWN
Location: /usr/local/lib/python2.7/dist-packages
Requires: backports.ssl-match-hostname, six, websocket-client, ipaddress, docker-pycreds, requests

ansible-playbook reports this:

ansible-playbook 2.2.0.0
  config file = /home/vagrant/Bitbucket/Operations/tlo-sa/installers/rpm/ansible.cfg
  configured module search path = Default w/o overrides

Hopefully, this can assist in tracking down the cause.

[solomongifford]

Interestingly, this is happening to us as well. I used the ansible playbooks last night (CentOS 6) with success but today it is failing. No yum updates show in the logs on the ansible machine. I'll update with anything I find.

[npetryk]

+1

Seeing this issue on new ubuntu 14.04.3 aws instances.

Ansible playbook == 2.2.0.0
Docker-py == 1.10.6
docker-engine == 1.12.3-0~trusty

Also hitting a private docker repository

[solomongifford]

@matrix10657 ours is a private docker repo as well.

[npetryk]

We're specifically using a quay.io repo. I suspect the problem is maybe with the repository considering that we're pinning all of the versions I mentioned, and we do a fresh install of all of them every time we build.

If we're all using quay then that's pretty suspicious.

[solomongifford]

@matrix10657 not using quay - and our private docker repo is is self hosted and did not have any package updates between the last successful run and the new failed runs.

[gbevan]

+1 same here.
Ubuntu 14.04.5 LTS
Kernel: 4.4.0-51
ansible: 2.2.0.0
docker-py: 1.9.0
Docker version 1.12.3, build 6b644ec

like others above, our builds have been working fine - tonight failing with this strange error. This was after ansible reported a successful image build then failed with this error attempting to start the image. docker images command does not show the supposedly newly built image name, instead shows:

REPOSITORY          TAG                 IMAGE ID            CREATED             SIZE
<none>              <none>              c2b5feecc16d        28 minutes ago      1.978 GB

[davegallant]

I think the issue is related to docker-compose. Switching down to 1.8.1 (on the target machine) fixed it:

- name: Install docker-compose
  pip: name=docker-compose version=1.8.1

[kurron]

I have docker-compose version 1.8.1, build 878cff1 installed on the failing box. Wait, @Davey-Dev , docker-compose is a Go program not a Python one so I don't see how how your fix works. What am I missing?

[rfrench-go]

I can confirm @Davey-Dev solution worked for me. Ansible playbooks were running just fine yesterday, but failed to start/restart containers today. I did NOT have docker-compose installed. I manually installed docker-compose on the target host:

pip install docker-compose==1.8.1

Immediately reran the playbook and my containers came up just fine.

AWS
Ubuntu 14.04
Ansbile 2.2.0
Docker-py 1.10.6

[davegallant]

@kurron I don't know much about Go, other than that Docker is written in Go and the 'http: no Host in request URL' error seems to be a specific to Go.

[MattAlex151]

I have the same issue with some writeup posted to SO.

EDIT: I did not previously have a task to install docker-compose, but I added one and installed 1.8.1 and can confirm this fixes the issue.

[kurron]

Obviously, I don't fully understand what is going on but running sudo pip install docker-compose==1.8.1 also fixed the issue for me. What I should have done was to interrogate pip before I upgraded to see what version of the Python module was installed. Could somebody run pip show docker-compose before upgrading so we can see what the differences in versions are?

[solomongifford]

@Davey-Dev docker-compose likely downgraded the requests module which is the actual culprit. The requests module came out with an update today (https://pypi.python.org/pypi/requests/2.12.2)

For anyone else, FYI, if your playbook is installing docker-py to the latest, then it is getting the latest version of requests. However, we actually need a lower version of requests.

I'm testing requests versions...

[solomongifford]

@kurron When I installed docker-compose, the requests module was the only one to update.

[ansibot]

@ansible, ping. This issue is waiting on your response.
click here for bot help

[kurron]

After successfully pulling down images again, I ran sudo pip install --upgrade docker-compose, which installed 1.9.0 onto the box and images seem to be coming down still.

Installing collected packages: requests, docker-compose
  Found existing installation: requests 2.7.0
    Uninstalling requests-2.7.0:
      Successfully uninstalled requests-2.7.0
  Found existing installation: docker-compose 1.8.1
    Uninstalling docker-compose-1.8.1:
      Successfully uninstalled docker-compose-1.8.1
  Running setup.py install for docker-compose ... done
Successfully installed docker-compose-1.9.0 requests-2.11.1

[solomongifford]

This issue should be closed in favor of the docker-py issue: docker/docker-py#1321