/docker-sqitch

Docker Image packaging for Sqitch

Primary LanguageDockerfile

Sqitch Docker Packaging

docker pull sqitch/sqitch
curl -L https://git.io/fAX6Z -o sqitch && chmod +x sqitch
./sqitch help

This project is the source for creating the official Sqitch Project Docker Image. It's built on stable Debian slim in an effort to keep the image as small as possible while supporting all known engines. It includes support for managing PostgreSQL, SQLite, MariaDB (MySQL), and Firebird databases, and other images may be built to support for the other database engines that Sqitch supports.

Notes

  • The docker-sqitch.sh shell script is the easiest way to run Sqitch from a Docker image. The script mounts the current directory and the home directory, so that it acts on the Sqitch project in the current directory and reads configuration from the home directory almost as if it was running natively on the local host. It also copies over most of the environment variables that Sqitch cares about, for transparent configuration.
  • By default, the container runs as the sqitch user. On macOS and Windows this works well, as files created by Sqitch in the working directory on the host will be properly owned by the host user. On Linux, however, docker-sqitch.sh runs the container with the UID and GID of the current host user, again so that files created by Sqitch will be owned by that user. If you find that the container cannot access configuration files in your home directory or write change scripts to the local directory, run sudo docker-sqitch.sh to run as the root user. Just be sure to chown files that Sqitch created for the consistency of your project.
  • If your engine falls back on the system username when connecting to the database (as the PostgreSQL engine does), you will likely want to set the username in sqitch target URIs, or set the proper environment variables to fall back on. Database authentication failures for the usernames sqitch or root are the hint you'll want to look for.
  • If you need to connect to a database server on your local host (or running in a container with the listening port mapped to the local host), use the host name host.docker.internal instead of localhost. Connections should work transparently when running Docker on Windows or macOS, although not yet on Linux (watch this PR for it to land). In the meantime you can use a NAT gateway container to forward traffic to the Docker host.
  • Custom images for Oracle, Snowflake, Exasol, or Vertica can be built by downloading the appropriate binary files and using the Dockerfiles in the appropriately-named subdirectories of this repository.
  • In an effort to keep things as simple as possible, the only editor included and configured for use in the image is nano. This is a very simple, tiny text editor suitable for editing change descriptions and the like. Its interface always provides menus to make it easy to figure out how to use it. If you need another editor, this image isn't for you, but you can create one based on this image and add whatever editors you like.