rind - Run tasks in dockerized environments, avoiding Docker in Docker (dind) and crazy workarounds.

• Run any command in docker or using the native environment
• Configurable log levels
• Structured outputs
• Terraform support

• bash 3+
• Docker

• Install on your target system and use the rind command:

  bash <(curl -sS https://raw.githubusercontent.com/timo-reymann/rind/main/installer)

• Download and directly pipe to bash

  curl -sS https://raw.githubusercontent.com/timo-reymann/rind/main/core/rind | bash -s - --log-level DEBUG

• Use the terraform module

Sometimes you have linux dependencies or tool chains you don't want to install locally but already have in your pipelines.

For example:

• python lambda that also uses deterministic-zip and includes dependencies that install native parts for linux
• you want to run a script that only works on *nix and don't want to adopt that it works on MacOS

You could run this locally on MacOS or even on a *nix-system and than would have problems with different target architectures.

So you could use always a docker container to run this tasks, but when you use CI you probably already use Docker for the job itself.

Using docker in docker is painful and not supported for all providers, so why not getting the best out of both worlds? - This is where rind comes in.

I love your input! I want to make contributing to this project as easy and transparent as possible, whether it's:

• Reporting a bug
• Discussing the current state of the configuration
• Submitting a fix
• Proposing new features
• Becoming a maintainer

To get started please read the Contribution Guidelines.

• GNU make
• Docker
• pre-commit

• Pure bash (posix compliant) + docker cli
• Decide based on /.dockerenv OR cgroup if we are running inside a container or not
• In container
  • Delete artifacts
  • Run setup
  • Run main
• On hosts
  • Copy current directory to build context
  • Delete artifacts to avoid readonly limitations
  • Run setup
  • Run main
  • Create image as volume container
  • Copy files to host

Copying files and small file changes are slow especially on MacOS. The ADD and COPY instructions are a LOT faster and workaround this problem.

This also allows docker layer cache to speed up your executions e.g. for terraform modules that build some code as part of the apply.

Docker in Docker is leading to a lot more problems and is a workaround itself. In most cases we already run in docker context due to an CI pipeline etc. So it makes sense to distinguish and make things easier and faster.

The project is split into multiple modules, each contains a guide on how to develop/test:

• core - Contains the core logic of rind
• terraform - Terraform module
• examples - Examples showcasing what you can do with rind
• installer - "curl-able" installer