tgiardina / docker-demo Goto Github PK

1. Things just work! No more worrying about dependencies and environments.
2. Will get you your next job (after you make your Datavant millions).

There are two main objects in Docker:

• Image - A snapshot of a Linux machine + a command to run on that machine.
• Container - A slice of your computer (i.e. disk space, RAM, etc) that runs an image.

We start by using the dev Dockerfile:

cp app/Dockerfile.dev app/Dockerfile

Build the image:

docker build app -t demo-app

app is the location of the application, which should be colocated with its Dockerfile. -t demo-app set the image name.

We can then run the container:

docker run -p 4000:3000 demo-app

-p 4000:3000 tells Docker to host on localhost:4000 whatever the container is hosting on port 3000. If you navigate to localhost:4000, you should see our app!

We will now use the prod Dockerfile:

cp app/Dockerfile.prod app/Dockerfile

Build the new image like we did last time and then run the container:

docker run -p 4000:80 demo-app

But we don't want our dev and prod setups diverging. E.g., if dev and prod start using different versions of node, then code that works in dev won't necessarily work on prod. Fortunately, we can combine the two Dockerfiles:

cp app/Dockerfile.final Dockefile

To build the prod env, use the standard:

docker build app -t demo-app

To build the dev env, specify a target stage:

docker build app --target installer -t demo-app-dev

To run the application, we now use:

docker run -p 4000:3000 demo-app-dev npm start

To run a container with a name:

docker run -p 4000:3000 --name dev-container demo-app-dev npm start

To see what containers are running, use:

docker ps

To start and stop the container, use:

docker start dev-container
docker stop dev-container

To run a command in a container, use:

docker exec -it dev-container <command>

To install packges within container:

apk update
apk add <package>

You can now take advantage of CRA's hot reloading using vim: vim src/App.js.

To clean up unused images nad containers:

docker system prune

To kill all containers:

docker kill $(docker ps -q)

To run a container and then detach, use the -d flag:

docker run -p 4000:3000 --name dev-container -d demo-app-dev npm start

Similarly, to start a container and attach to it use the -a flage:

docker start -a dev-container

And if you're running a detached container, you can reattach:

docker attach dev-container

But a React app by itself isn't that useful. We probably want some sort of API and database as well. And it's a lot of work to spin those all up independently. docker-compose let's us coordinate all these containers in one file (similar to k8s)!

Use the prod docker-compose.yaml:

cp docker-compose.prod.yaml docker-compose.yaml

Start up the containers using:

docker-compose up --build

where --build ensures we're working with up-to-date images.

Same as prod, but use docker-compose.dev.yaml instead:

cp docker-compose.dev.yaml docker-compose.yaml

Again, we don't want our dev and prod envs diverging, so we'd like to reuse as much config as possible. Start by using the combined docker compose file:

cp docker-compose.final.yaml docker-compose.yaml

Then to spin up prod, use the prod .env:

cp .env.prod .env
docker-compose up --build

And similarly for dev .env:

cp .env.dev .env
docker-compose up --build

• This Udemy course is good (Thanks Matt!).