skullydazed's dotfiles

Installation

You can install this with:

curl https://raw.githubusercontent.com/skullydazed/dotfiles/main/install.sh | bash

It will update itself every time you login.

Things to know about

.note files

Put a file named .note into a directory and the contents will be displayed every time you enter that directory.

Updating

When you login the dotfiles will be updated automatically in the background, and any missing symlinks will be added. If you change where a symlink points to the update system will not change it- this gives you a way to handle local overrides if needed.

Your .profile changes will not be picked up until the next time you login.

Docker

You can use docker_deploy to deploy the docker service in the current directory to your local machine. It will build the service, tag it after the directory name, and launch it. You must create a DOCKER_RUN file to control the behavior of your service.

The following docker run arguments will always be passed by docker_deploy:

-it --name $container_name --restart=unless-stopped

You can specify more arguments for both docker run and the container's entrypoint in the DOCKER_RUN file. DOCKER_RUN will be sourced by docker_deploy to specify these arguments.

Variable	Description
`DOCKER_RUN_ARGS`	Arguments to pass to `docker run`, such as `--port`
`DOCKER_ENTRYPOINT_ARGS`	Arguments to pass to the container

Docker Deploy Example

You can run the following commands to create a minimal docker_deploy directory:

mkdir hello-world
cd hello-world
echo 'FROM hello-world' > Dockerfile
echo 'DOCKER_RUN_ARGS=""' > DOCKER_RUN 
docker_deploy

This will build and tag a hello-world container and launch it, leaving you connected to the console.

Why not use <docker-compose|swarm|k8s|rancher> instead?

For application development purposes docker-compose is great. When it comes to deploying in a production cluster you'll setup a lot of infrastructure around Rancher or Kubernetes. But if you have only one server that you want to deploy several docker containers to and manage them semi-independently you'll run into friction using docker-compose, which tries to treat all your containers as being part of larger service. The best way to go for a standalone production server is to setup Rancher or k8s, but they both consume a pretty heafty chunk of your time setting them up and can eat up 2-4gb of your available RAM.

The nice thing about docker_deploy is that all you need to use it is a Dockerfile and DOCKER_RUN in a directory dedicated to a specific container. You can manage each container much like you manage the services on a machine.

skullydazed / dotfiles