Docker containers running Alpine Linux and s6 for process management. Solid, reliable containers.
Highly configurable Docker images running Alpine linux and s6 process management.
Table of contents
This project has the following goals:
To meet these goals, we’re using:
Using Docker makes your infrastructure and environment consistent, testable, scalable and repeatable.
Microservices are isolated components (of a whole) doing one thing, and doing it well. These microservices are pulled together to provide a complete platform. For example, each of the following microservices would exist within one Docker container:
That’s called service discovery. We solve that problem using Consul (which is a distributed service discovery tool, with built-in health checks). Each container advertises their service (IP and port) via Consul.
Each container with a service, connects to a member of the Consul cluster. Consul then talks to all members to enable service discovery via HTTP and DNS.
A good example is a container with Nginx that will query Consul for the IP address of the Node.js container so that it can proxy an incoming request through to the Node.js application.
Setting up your infrastructure using microservices makes scalability a breeze.
There is an example showing Nginx load balancing incoming requests to multiple Node.js containers.
These images are used heavily in production by a number of different companies. They’ve been designed as great starting points to easily customize and use within your own platforms.
The following are common to all images within this repository.
Alpine Linux provides the foundation for all images in this repository. It’s small, fast and perfect for Docker.
s6 is a unix service supervisor (much like runit and supervisord only better).
While it’s true that Docker containers should have one clear focus, more often than not, you’ll need to run multiple processes within the container.
Consul is a great example; you need the Consul agent to join a Consul cluster and the primary service such as Nginx or Node.js. Log management is also a key consideration in which another process might need to run.
s6-overlay is s6 Dockerized! It makes working with Docker and process management via s6 super easy. It also features some really nice extras such as container initialization and finalization stages (run custom scripts during these phases to setup services or tidy up from crashes).
go-dnsmasq has been added into the mix to add support for the DNS search keyword which doesn’t come standard in Alpine Linux.
These images work well with the embedded DNS service that comes with Docker 1.10+.
Configuration is a key step within a Docker-based microservices architecture. You must configure your services to interact with each other.
All Consul-based images come with consul-template for easy configuration based on changes within Consul.
All non-Consul-based images come with confd.
All Consul-related containers have been design for use in production and development.
These containers have been configured for zero-conf Consul bootstrapping. To achieve this, docker-engine v1.10+ is required. along with a docker-compose.yml file written in the version 2 format (see the example).
To achieve zero-conf Consul bootstrapping, Docker’s new embedded DNS server is used within each container to find the IP of a consul container (note: there is a requirement that all Consul server services in docker-compose.yml are called consul for this to work).
Note:: this can work in older versions of Docker however (using service links with the name consul).
The Consul-based images in this repository have been designed to meet the following requirements:
Containers running Consul in server mode:
Containers running Consul in agent mode:
We’ve found this to be the most consistent way to effectively run Consul in production. This also supports the development workflow in which you CTRL+C to stop a container from running.
Most images have an example in the examples folder.
There is also a complete example which demonstrates:
You can read more about the example.
The following describes the images that are available and the inheritance chain.
└─ alpine-base
├─ alpine-apache
├─ alpine-confd
| └─ alpine-rabbitmq
├─ alpine-nginx
| └─ alpine-nginx-nodejs
├─ alpine-nodejs
└─ alpine-redis
└─ alpine-base
└─ alpine-consul
├─ alpine-consul-ui
└─ alpine-consul-base
├─ alpine-consul-apache
├─ alpine-consul-nodejs
├─ alpine-consul-nginx
| └─ alpine-consul-nginx-nodejs
└─ alpine-consul-redis
This image is the base for all containers. All other Docker images within this repository inherit from this Container.
Latest version is 3.3.0, or latest.
This image includes Apache HTTPD with a very basic configuration.
Latest version is 2.0.1, or latest.
This image adds confd. It should be seen as a base image suitable for heavy customisation.
Latest version is 3.1.0, or latest.
This image adds Consul.
If you want to create a Docker image to run as a Consul agent in server mode (i.e. part of a cluster) start with this image.
Latest version is 3.2.0, or latest.
This image inherits from alpine-consul and is designed as a base image for other Docker images which will join to a Consul cluster.
If you want to create a Docker image to advertise a service in Consul start with this image.
Latest version is 4.2.0, or latest.
This image is designed to run Apache within the context of service discovery (via Consul).
It is suited to running Apache as a proxy to another Docker container.
Latest version is 2.0.0, or latest.
This image is designed to run a Node.js application within the context of service discovery (via Consul).
Latest version is 5.11.0, or latest.
This image is designed to run Nginx within the context of service discovery (via Consul).
It’s suited for running an Nginx proxy or to load balance with Nginx to another container.
Latest version is 3.2.0, or latest.
This image is designed to run both Nginx and Node.js on the same container. It’s a shortcut to running alpine-consul-nodejs via a proxy from alpine-consul-nodejs.
Latest version is 2.0.0, or latest.
This image has been designed to run Redis within the context of service discovery (via Consul).
Latest version is 2.1.0, or latest.
This image has been designed to connect to a Consul cluster (from alpine-consul) and make the Consul admin UI accessible.
Latest version is 2.2.0, or latest.
This image includes Nginx with a very basic setup.
Latest version is 3.0.0, or latest.
This image includes both Nginx and Node.js. It’s suitable if you want to have Node.js perform configuration for Nginx.
Latest version is 4.4.0, or latest.
This includes Node.js.
Latest version is 8.15.0, or latest.
This image inherits from alpine-confd, and includes RabbitMQ with a basic setup.
Latest version is 2.1.1, or latest.
This image includes Redis.
Latest version is 1.0.0, or latest.
All containers come with the following features (most of them come from s6-overlay):
CMDAs a general rule, these images eschew the CMD option that Dockerfile provides. These images are oriented towards running multiple services.
However, alpine-base is a minimalist image that can be used with CMD. Your command can be provided in the Dockerfile or at runtime and it will be run under the s6 supervisor. When the command fails or exits, the container will also exit (falling back to Docker restart rules or other platform controls to manage container availability).
For example:
vagrant@ubuntu-14:/vagrant$ docker run -it smebberson/alpine-base with-contenv sh
[s6-init] making user provided files available at /var/run/s6/etc...exited 0.
[s6-init] ensuring user provided files have correct perms...exited 0.
[fix-attrs.d] applying ownership & permissions fixes...
[fix-attrs.d] done.
[cont-init.d] executing container initialization scripts...
[cont-init.d] 30-resolver: executing...
[cont-init.d] 30-resolver: exited 0.
[cont-init.d] 40-resolver: executing...
[cont-init.d] 40-resolver: exited 0.
[cont-init.d] done.
[services.d] starting services
[services.d] done.
/ # env
HOSTNAME=63d07768fdc5
SHLVL=1
HOME=/root
GO_DNSMASQ_LOG_FILE=/var/log/go-dnsmasq/go-dnsmasq.log
GODNSMASQ_VERSION=1.0.5
TERM=xterm
S6_OVERLAY_VERSION=v1.19.1.1
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
PWD=/
/ # exit
with-contenv exited 0
[cont-finish.d] executing container finish scripts...
[cont-finish.d] done.
[s6-finish] syncing disks.
[s6-finish] sending all processes the TERM signal.
[s6-finish] sending all processes the KILL signal and exiting.
vagrant@ubuntu-14:/vagrant$
s6 provides us a clean and simple way to run services within these images. For example alpine-consul runs both go-dnsmasq and Consul via services, and alpine-consul-nginx adds Nginx, also via services.
To write a service script, create a directory with the name of your service in the /etc/services.d/ directory. In that directory create an executable script named run. In your run script, start your service (be sure to make sure it doesn’t background as s6 will attempt to start it again). For example:
# /etc/services.d/nginx/run
#!/usr/bin/with-contenv sh
nginx -g "daemon off;"
s6 will start your service and monitor it. Should your service exit or fail, s6 will execute your run script again to restart the service.
s6 also provides you a hook to run tasks in between a service exiting or failing, and it being restarted. Simple create an executable script named finish next to your run script.
This script should execute fast. It has about 3 seconds to do what it needs to before it will be killed by s6, and the run script executed. You can also do any tasks as required in your run before you start your long running process.
If you’d like to bring down a container if a particular service restarts, use this as the contents of a finish script:
#!/usr/bin/execlineb -S1
# only tell s6 to bring down the entire container, if it isn't already doing so
# http://skarnet.org/software/s6/s6-supervise.html
if { s6-test ${1} -ne 0 }
if { s6-test ${1} -ne 256 }
s6-svscanctl -t /var/run/s6/services
The docker-alpine images are highly customizable. You can customize them with the following:
They’ve also been designed so that it’s easy to use these images in a non-standard context such as Docker Cloud and Rancher.
The following cover customization that is common to all images. Review the documentation for each image for image-specific customization options.
/usr/bin/host-ipThis file is used in other scripts to determine the IP of the container in which a script is running.
Overwrite it as required to produce a correct value for your environment. Review the file itself for output requirements.
If your environment uses an overlay network (Docker Cloud, Rancher), you should change this file.
/usr/bin/container-findThis file (Consul-only containers) is used within other scripts to determine the IP address(es) of other containers by name.
By default, it uses dig to DNS query for the IP address of a container by name. container-find will DNS query for consul (the default). container-find static will DNS query for static. container-find static.service.consul will hand-off the DNS query to Consul.
Each image has it’s own version and will be updated independently. All versions follow semver. If a software package or the operating system the image uses is updated, the level of update will be reflected in the new version number, for example:
These images are a little different from your standard Docker images. The following should explain most of the differences.
By default, Alpine Linux <= v3.4 doesn’t support DNS search. This has been enabled through the use of go-dnsmasq.
These images don’t contain Bash by default. They’re defaulted to use sh.
You can either write your scripts such that they’re posix compliant or install Bash with apk add --no-cache bash.
By default, sh has no environment. If you need access to environment variables in scripts or when using docker exec you should use:
#!/usr/bin/with-contenv shdocker exec -it complete_consul_1 with-contenvThere are two ways to set environment variables. In your Dockerfile like any other containers or from within a custom script using set-contenv for example set-contenv VAR_NAME var_value.
If you’d like to read more about operating environment of these images, start here:
Read the CHANGELOG.md for a chronological record of events. Each image also has it’s own CHANGELOG.
We love contributors. Read CONTRIBUTING.md for more information on how to contribute to this project.
You can view more information about the contributors here.