Docker

From Christoph's Personal Wiki
Revision as of 19:23, 8 March 2018 by Christoph (Talk | contribs)

Jump to: navigation, search

Docker is an open-source project that automates the deployment of applications inside software containers. Quote of features from docker web page:

Docker containers wrap up a piece of software in a complete filesystem that contains everything it needs to run: code, runtime, system tools, system libraries – anything you can install on a server. This guarantees that it will always run the same, regardless of the environment it is running in.[1]

Dockerfile directives

USER

$ cat << EOF > Dockerfile
# Non-privileged user entry
FROM centos:latest
MAINTAINER xtof@example.com

RUN useradd -ms /bin/bash xtof
USER xtof
EOF

Note: The use of MAINTAINER has been deprecated in newer versions of Docker. You should use LABEL instead, as it is much more flexible and its key/values show up in docker inspect. From here forward, I will only use LABEL.

$ docker build -t centos7/nonroot:v1 .
$ docker exec -it <container_name> /bin/bash

We are user "xtof" and are unable to become root. The workaround (i.e., how to become root) is like so:

$ docker exec -u 0 -it <container_name> /bin/bash

NOTE: For the remainder of this section, I will omit the $ cat << EOF > Dockerfile part in the examples for brevity.

RUN

Notes on the order of execution

FROM centos:latest
LABEL maintainer="xtof@example.com"

RUN useradd -ms /bin/bash xtof
USER xtof

RUN echo "export PATH=/path/to/my/app:$PATH" >> /etc/bashrc
$ docker build -t centos7/config:v1 .
...
/bin/sh: /etc/bashrc: Permission denied

The order of execution matters! Prior to the directive USER xtof, the user was root. After that directive, the user is now xtof, who does not have super-user privileges. Move the RUN echo ... directive to before the USER xtof directive for a successful build.

ENV

Note: The following is a _terrible_ way of building a container. I am purposely doing it this way so I can show you a much better way later (see below).

  • Build a CentOS 7 Docker image with Java 8 installed:
# SEE: https://gist.github.com/P7h/9741922 for various Java versions
FROM centos:latest
LABEL maintainer="xtof@example.com"

RUN yum update -y
RUN yum install -y net-tools wget

RUN echo "SETTING UP JAVA"
# The tarball method:
#RUN cd ~ && wget --no-cookies --no-check-certificate \
#    --header "Cookie: gpw_e24=http%3A%2F%2Fwww.oracle.com%2F; oraclelicense=accept-securebackup-cookie" \
#    "http://download.oracle.com/otn-pub/java/jdk/8u91-b14/jdk-8u91-linux-x64.tar.gz"
#RUN tar xzvf jdk-8u91-linux-x64.tar.gz
#RUN mv jdk1.8.0_91 /opt
#ENV JAVA_HOME /opt/jdk1.8.0_91/

# The rpm method:
RUN cd ~ && wget --no-cookies --no-check-certificate \
    --header "Cookie: gpw_e24=http%3A%2F%2Fwww.oracle.com%2F; oraclelicense=accept-securebackup-cookie" \
    "http://download.oracle.com/otn-pub/java/jdk/8u161-b12/2f38c3b165be4555a1fa6e98c45e0808/jdk-8u161-linux-x64.rpm"
RUN yum localinstall -y /root/jdk-8u161-linux-x64.rpm

RUN useradd -ms /bin/bash xtof
USER xtof

# User specific environment variable
RUN cd ~ && echo "export JAVA_HOME=/usr/java/jdk1.8.0_161/jre" >> ~/.bashrc
# Global (system-wide) environment variable
ENV JAVA_BIN /usr/java/jdk1.8.0_161/jre/bin
$ docker build -t centos7/java8:v1 .

CMD vs. RUN

FROM centos:latest
LABEL maintainer="xtof@example.com"

RUN useradd -ms /bin/bash xtof
CMD ["echo", "Hello from within my container"]

The CMD directive only executes when the container is started, whereas the RUN directive is executed during the build of the image.

$ docker build -t centos7/echo:v1 .
$ docker run centos7/echo:v1
Hello from within my container

The container starts, echos out that message, then exits.

ENTRYPOINT

FROM centos:latest
LABEL maintainer="xtof@example.com"

RUN useradd -ms /bin/bash xtof
ENTRYPOINT "This command will display this message on EVERY container that is run from it"
$ docker build -t centos7/entry:v1 .
$ docker run centos7/entry:v1
This command will display this message on EVERY container that is run from it
$ docker run centos7/entry:v1 /bin/echo "Can you see me?"
This command will display this message on EVERY container that is run from it
$ docker run centos7/echo:v1 /bin/echo "Can you see me?"
Can you see me?

Note the difference.

EXPOSE

FROM centos:latest
LABEL maintainer="xtof@example.com"

RUN yum update -y
RUN yum install -y httpd net-tools

RUN echo "This is a custom index file built during the image creation" > /var/www/html/index.html

ENTRYPOINT apachectl -DFOREGROUND  # BAD WAY TO DO THIS!
$ docker build -t centos7/apache:v1 .
$ docker run -d --name webserver centos7/apache:v1
$ docker exec webserver /bin/cat /var/www/html/index.html
This is a custom index file built during the image creation
$ docker inspect webserver -f '{{.NetworkSettings.IPAddress}}'  # => 172.17.0.6
#~OR~
$ docker inspect webserver | jq -crM '.[] | .NetworkSettings.IPAddress'  # => 172.17.0.6
$ curl 172.17.0.6
This is a custom index file built during the image creation
$ curl -sI 172.17.0.6 | awk '/^HTTP|^Server/{print}'
HTTP/1.1 200 OK
Server: Apache/2.4.6 (CentOS)
$ time docker stop webserver
real	0m10.275s  # <- notice how long it took to stop the container
user	0m0.008s
sys	0m0.000s
$ docker rm webserver

It took ~10 seconds to stop the above container. This is because of the way we are (incorrectly) using ENTRYPOINT. The SIGTERM signal when running `docker stop webserver` actually timed out instead of exiting gracefully. A much better method is show below, which will exit gracefully and in less than 300 ms.

  • Expose ports from the CLI
$ docker run -d --name webserver -p 8080:80 centos7/apache:v1
$ curl localhost:8080
This is a custom index file built during the image creation
$ docker stop webserver && docker rm webserver
  • Explicitly expose a port in the Docker image:
FROM centos:latest
LABEL maintainer="xtof@example.com"

RUN yum update -y
RUN yum install -y httpd net-tools

RUN echo "This is a custom index file built during the image creation" > /var/www/html/index.html

EXPOSE 80

ENTRYPOINT ["/usr/sbin/httpd", "-D", "FOREGROUND"]
$ docker build -t centos7/apache:v1 .
$ docker run -d --rm --name webserver -P centos7/apache:v1
$ docker container ls --format '{{.Names}} {{.Ports}}'
webserver 0.0.0.0:32769->80/tcp
#~OR~
$ docker inspect webserver | jq -crM '[.[] | .NetworkSettings.Ports."80/tcp"[] | .HostPort] | .[]'
32769
$ curl localhost:32769
This is a custom index file built during the image creation
$ time docker stop webserver
real	0m0.283s
user	0m0.004s
sys	0m0.008s

Note that I passed --rm to the `docker run` command so that the container will be removed when I stop the container. Also note how much faster the container stopped (~300ms vs. 10 seconds above).

Container volume management

$ docker run -it --name voltest1 -v /mydata centos:latest /bin/bash
[root@bffdcb88c485 /]# df -h
Filesystem                   Size  Used Avail Use% Mounted on
none                         213G  173G   30G  86% /
tmpfs                        7.8G     0  7.8G   0% /dev
tmpfs                        7.8G     0  7.8G   0% /sys/fs/cgroup
/dev/mapper/ubuntu--vg-root  213G  173G   30G  86% /mydata
shm                           64M     0   64M   0% /dev/shm
tmpfs                        7.8G     0  7.8G   0% /sys/firmware
[root@bffdcb88c485 /]# echo "testing" >/mydata/mytext.txt
$ docker inspect voltest1 | jq -crM '.[] | .Mounts[].Source'
/var/lib/docker/volumes/2a53fd295595690200a63def8a333b54682174923339130d560fb77ecbe41a3b/_data
$ sudo cat /var/lib/docker/volumes/2a53fd295595690200a63def8a333b54682174923339130d560fb77ecbe41a3b/_data/mytext.txt
testing
$ sudo echo "this is from the host OS" >/var/lib/docker/volumes/2a53fd295595690200a63def8a333b54682174923339130d560fb77ecbe41a3b/_data/host.txt
[root@bffdcb88c485 /]# cat /mydata/host.txt 
this is from the host OS

Install docker

Debian-based distros

Note: For this install, I will be using Ubuntu 16.04 LTS (Xenial Xerus). Docker requires a 64-bit version of Ubuntu as well as a kernel version equal to or greater than 3.10. My system satisfies both requirements.

  • Setup the docker repo to install from:
$ sudo apt-get update -y
$ sudo apt-key adv --keyserver hkp://p80.pool.sks-keyservers.net:80 --recv-keys 58118E89F3A912897C070ADBF76221572C52609D
$ echo "deb https://apt.dockerproject.org/repo ubuntu-xenial main" | sudo tee /etc/apt/sources.list.d/docker.list
$ sudo apt-get update -y

Make sure you are about to install from the Docker repo instead of the default Ubuntu 16.04 repo:

$ apt-cache policy docker-engine

The output of the above command show look something like the following:

docker-engine:
  Installed: (none)
  Candidate: 1.11.2-0~xenial
  Version table:
     1.11.2-0~xenial 500
        500 https://apt.dockerproject.org/repo ubuntu-xenial/main amd64 Packages
     1.11.1-0~xenial 500
        500 https://apt.dockerproject.org/repo ubuntu-xenial/main amd64 Packages
     1.11.0-0~xenial 500
        500 https://apt.dockerproject.org/repo ubuntu-xenial/main amd64 Packages
  • Install docker:
$ sudo apt-get install -y docker-engine

Red Hat-based distros

Note: For this install, I will be using CentOS 7 (release 7.2.1511). Docker requires a 64-bit version of CentOS as well as a kernel version equal to or greater than 3.10. My system satisfies both requirements.

  • Install Docker (the fast way):
$ sudo yum update -y
$ curl -fsSL https://get.docker.com/ | sh
  • Install Docker (via a yum repo):
$ sudo yum update -y
$ sudo pip install docker-py
$ cat << EOF > /etc/yum.repos.d/docker.repo
[dockerrepo]
name=Docker Repository
baseurl=https://yum.dockerproject.org/repo/main/centos/7/
enabled=1
gpgcheck=1
gpgkey=https://yum.dockerproject.org/gpg
EOF
$ sudo rpm -vv --import https://yum.dockerproject.org/gpg
$ sudo yum update -y
$ sudo yum install docker-engine -y

Post-installation steps

  • Check on the status of docker:
$ sudo systemctl status docker
● docker.service - Docker Application Container Engine
   Loaded: loaded (/lib/systemd/system/docker.service; enabled; vendor preset: enabled)
   Active: active (running) since Tue 2016-07-12 12:31:08 PDT; 6s ago
     Docs: https://docs.docker.com
 Main PID: 3392 (docker)
   CGroup: /system.slice/docker.service
           ├─3392 /usr/bin/docker daemon -H fd://
           └─3411 docker-containerd -l /var/run/docker/libcontainerd/docker-containerd.sock --runtime docker-runc --start-timeout 2m
  • Make sure the docker service automatically starts after a machine reboot:
$ sudo systemctl enable docker
  • Execute docker without `sudo`:
$ sudo usermod -aG docker $(whoami)

Log out and log back in to use docker without `sudo`.

  • Check that docker has been successfully installed and configured:
$ docker run hello-world
...
This message shows that your installation appears to be working correctly.
...

Install your own Docker private registry

Note: I will use CentOS 7 for this install and assume you already have docker and docker-compose installed (see above).

For this install, I will assume you have a domain name registered somewhere. I will use docker.example.com as my example domain. Replace anywhere you see that below with your actual domain name.

  • Install dependencies:
$ yum install -y nginx  # used for the registry endpoint
$ yum install -y httpd-tools  # for the htpasswd utility
  • Setup docker registry directory structure:
$ mkdir -p /opt/docker-registry/{data,nginx{/conf.d,/certs},log}
$ cd /opt/docker-registry
  • Create a docker-compose file:
$ vim docker-compose.yml  # and add the following:
nginx:
  image: "nginx:1.9"
  ports:
    - 5043:443
  links:
    - registry:registry
  volumes:
    - ./log/nginx/:/var/log/nginx:rw
    - ./nginx/conf.d:/etc/nginx/conf.d:ro
    - ./nginx/certs:/etc/nginx/certs:ro
registry:
  image: registry:2
  ports:
    - 127.0.0.1:5000:5000
  environment:
    REGISTRY_STORAGE_FILESYSTEM_ROOTDIRECTORY: /data
  volumes:
    - ./data:/data
  • Create an Nginx configuration file:
$ vim /opt/docker-registry/nginx/conf.d/registry.conf  # and add the following:
upstream docker-registry {
  server registry:5000;
}

server {
  listen 443;
  server_name docker.example.com;

  # SSL
  ssl on;
  ssl_certificate /etc/nginx/certs/docker.example.com.crt;
  ssl_certificate_key /etc/nginx/certs/docker.example.com.key;

  # disable any limits to avoid HTTP 413 for large image uploads
  client_max_body_size 0;

  # required to avoid HTTP 411: see Issue #1486 (https://github.com/docker/docker/issues/1486)
  chunked_transfer_encoding on;

  location /v2/ {
    # Do not allow connections from docker 1.5 and earlier
    # docker pre-1.6.0 did not properly set the user agent on ping, catch "Go *" user agents
    if ($http_user_agent ~ "^(docker\/1\.(3|4|5(?!\.[0-9]-dev))|Go ).*$" ) {
      return 404;
    }

    proxy_pass                          http://docker-registry;
    proxy_set_header  Host              $http_host;   # required for docker client's sake
    proxy_set_header  X-Real-IP         $remote_addr; # pass on real client's IP
    proxy_set_header  X-Forwarded-For   $proxy_add_x_forwarded_for;
    proxy_set_header  X-Forwarded-Proto $scheme;
    proxy_read_timeout                  900;

    add_header 'Docker-Distribution-Api-Version:' 'registry/2.0' always;

    # To add basic authentication to v2 use auth_basic setting plus add_header
    auth_basic "Restricted access to Docker Registry";
    auth_basic_user_file /etc/nginx/conf.d/registry.htpasswd;
  }
}
$ cd /opt/docker-registry/nginx/conf.d
$ htpasswd -c registry.htpasswd <username>  # replace <username> with your actual username
$ htpasswd registry.htpasswd <username2>  # [optional] add a 2nd user
  • Setup your own certificate signing authority (for use with SSL):
$ cd /opt/docker-registry/nginx/certs
  • Generate a new root key:
$ openssl genrsa -out docker-registry-CA.key 2048
  • Generate a root certificate (enter anything you like at the prompts):
$ openssl req -x509 -new -nodes -key docker-registry-CA.key -days 3650 -out docker-registry-CA.crt

Then generate a key for your server (this is the file referenced by ssl_certificate_key in the Nginx configuration above):

$ openssl genrsa -out docker.example.com.key 2048

Now we have to make a certificate signing request (CSR). After you type the following command, OpenSSL will prompt you to answer a few questions. Enter anything you like for the first few, however, when OpenSSL prompts you to enter the "Common Name", make sure to enter the domain or IP of your server.

$ openssl req -new -key docker.example.com.key -out docker.example.com.csr
  • Sign the certificate request:
$ openssl x509 -req -in docker.example.com.csr -CA docker-registry-CA.crt -CAkey docker-registry-CA.key -CAcreateserial -out docker.example.com.crt -days 3650
  • Force any clients that will use the certificate authority we created above to accept that it is a "legitimate" certificate. Run the following commands on the Docker registry server and on any hosts that will be communicating with the Docker registry server:
$ sudo cp /opt/docker-registry/nginx/certs/docker-registry-CA.crt /usr/local/share/ca-certificates/
$ sudo update-ca-trust
  • Restart the Docker daemon in order for it to pick up the changes to the certificate store:
$ sudo systemctl restart docker.service
  • Bring up the associated Docker containers:
$ docker-compose up -d
  • Your Docker registry directory structure should look like the following:
$ cd /opt/docker-registry && tree .
.
├── data
├── docker-compose.yml
├── log
│   └── nginx
│       ├── access.log
│       └── error.log
└── nginx
    ├── certs
    │   ├── docker-registry-CA.crt
    │   ├── docker-registry-CA.key
    │   ├── docker-registry-CA.srl
    │   ├── docker.example.com.crt
    │   ├── docker.example.com.csr
    │   └── docker.example.com.key
    └── conf.d
        ├── registry.conf
        └── registry.htpasswd
  • To access the private Docker registry from a client machine (any machine, really), first add the SSL certificate you created earlier to the client machine:
$ cat /opt/docker-registry/nginx/certs/docker-registry-CA.crt  # copy contents
# On client machine:
$ sudo vim /usr/local/share/ca-certificates/docker-registry-CA.crt  # paste contents
$ sudo update-ca-certificates  # You should see "1 added" in the output
  • Restart Docker on the client machine to make sure it reloads the system's CA certificates:
$ sudo service docker restart
  • Test that you can reach your private Docker registry:
$ curl -k https://USERNAME:PASSWORD@docker.example.com:5043/v2/
{}  # <- proper output
  • Now, test that you can login with Docker:
$ docker login https://docker.example.com:5043

If that returns with "Login Succeeded", your private Docker registry is up and running!

This section is incomplete. It will be updated presently.

Docker environment variables

Note: See here for the most up-to-date list of environment variables.

The following list of environment variables are supported by the docker command line:

DOCKER_API_VERSION 
The API version to use (e.g., 1.19)
DOCKER_CONFIG 
The location of your client configuration files.
DOCKER_CERT_PATH 
The location of your authentication keys.
DOCKER_DRIVER 
The graph driver to use.
DOCKER_HOST 
Daemon socket to connect to.
DOCKER_NOWARN_KERNEL_VERSION 
Prevent warnings that your Linux kernel is unsuitable for Docker.
DOCKER_RAMDISK 
If set this will disable "pivot_root".
DOCKER_TLS_VERIFY 
When set Docker uses TLS and verifies the remote.
DOCKER_CONTENT_TRUST 
When set Docker uses notary to sign and verify images. Equates to --disable-content-trust=false for build, create, pull, push, run.
DOCKER_CONTENT_TRUST_SERVER 
The URL of the Notary server to use. This defaults to the same URL as the registry.
DOCKER_TMPDIR 
Location for temporary Docker files.

Because Docker is developed using "Go", one can also use any environment variables used by the "Go" runtime. In particular, the following might be useful:

HTTP_PROXY
HTTPS_PROXY
NO_PROXY
  • Example usage:
$ export DOCKER_API_VERSION=1.19

References

  1. https://www.docker.com/what-docker

External links