Skip to content

Latest commit

 

History

History
113 lines (78 loc) · 5.49 KB

docker.md

File metadata and controls

113 lines (78 loc) · 5.49 KB
Raw

Docker Support

The inclusion of a Dockerfile in the dcrdata repository means you can use Docker for dcrdata development or in production. However, official images are not presently published to docker hub.

When developing you can utilize containers for easily swapping out Go versions and overall project setup. You don't even need go installed on your system if using containers during development.

Once Docker is installed, you can then download this repository and follow the build instructions below.

Building the Image

To use a dcrdata container you need to build an image as follows:

docker build --squash -t decred/dcrdata:dev-alpine .

Note: The --squash flag is an experimental feature as of Docker 18.06. Experimental features must be enabled to use the setting. On Windows and OS/X, look under the "Daemon" settings tab. On Linux, enable the setting manually.

By default, docker will build the container based on the Dockerfile found in the root of the repository that is based on Alpine Linux. To use an Ubuntu-based container, you should build from the Ubuntu-based Dockerfile:

docker build --squash -f dockerfiles/Dockerfile_stretch -t decred/dcrdata:dev-stretch .

Part of the build process is to copy all the source code over to the image, download all dependencies, and build dcrdata. If you run into build errors with docker try adding the --no-cache flag to trigger a rebuild of all the layers since docker does not rebuild cached layers.

docker build --no-cache --squash -t decred/dcrdata:dev-alpine .

Building dcrdata with Docker

In addition to running dcrdata in a container, you can also build dcrdata inside a container and copy the executable to another system. To do this, you must have the dcrdata Docker image or build it from source.

The default container image is based on amd64 Alpine Linux. To create a binary targeting different operating systems or architectures, it is necessary to set the GOOS and GOARCH environment variables.

From the repository source folder, do the following to build the Docker image, and compile dcrdata into your current directory:

  • docker build --squash -t decred/dcrdata:dev-alpine . Only build the container image if necessary
  • docker run --entrypoint="" -v ${PWD}:/home/decred/go/src/github.com/decred/dcrdata --rm decred/dcrdata:dev-alpine go build

This mounts your current working directory in the host machine on a volume inside the container so that the build output will be on the host file system.

Build for other platforms as follows:

docker run -e GOOS=darwin -e GOARCH=amd64 --entrypoint="" -v ${PWD}:/home/decred/go/src/github.com/decred/dcrdata --rm decred/dcrdata:dev-alpine go build

docker run -e GOOS=windows -e GOARCH=amd64 --entrypoint="" -v ${PWD}:/home/decred/go/src/github.com/decred/dcrdata --rm decred/dcrdata:dev-alpine go build

Developing dcrdata Using a Container

Containers are a great way to develop any source code as they serve as a disposable runtime environment built specifically to the specifications of the application. Suggestions for developing in a container:

  1. Don't write code inside the container.
  2. Attach a volume and write code from your editor on your docker host.
  3. Attached volumes on a Mac are generally slower than Linux/Windows.
  4. Install everything in the container, don't muck up your Docker host.
  5. Resist the urge to run git commands from the container.
  6. You can swap out the Go version just by using a different docker image.

To make the source code from the host available inside the container, attach a volume to the container when launching the image:

docker run -ti --entrypoint="" -v ${PWD}:/home/decred/go/src/github.com/decred/dcrdata --rm decred/dcrdata:dev-alpine /bin/bash

Note: Changing entrypoint allows you to run commands in the container since the default container command runs dcrdata. We also added /bin/bash at the end so the container executes this by default.

You can now run go build or go test inside the container. If you run go fmt you should notice that any formatting changes will also be reflected on the docker host as well.

To run dcrdata in the container, it may be convenient to use environment variables to configure dcrdata. The variables may be set inside the container or on the command line. For example,

docker run -ti --entrypoint=/bin/bash -e DCRDATA_LISTEN_URL=0.0.0.0:2222 -v ${PWD}:/home/decred/go/src/github.com/decred/dcrdata --rm decred/dcrdata:dev-alpine

Container Production Usage

We don't yet have a build system in place for creating production grade images of dcrdata. However, you can still use the images for testing.

In addition to configuring dcrdata, it is also necessary to map the TCP port on which dcrdata listens for connections with the -p switch. For example,

docker run -ti -p 2222:2222 -e DCRDATA_LISTEN_URL=0.0.0.0:2222 --rm decred/dcrdata:dev-alpine

Please keep in mind these images have not been hardened so this is not recommended for production.

Note: The TLS certificate for dcrd's RPC server may be needed in the container. Either build a new container image with the certificate, or attach a volume containing the certificate to the container.