angelospanag / fastapi-podman

An example project using Podman to run a container with Python/FastAPI to create a web server and Poetry for dependency management.

fastapi-podman

An example project using Podman to run a container with Python/FastAPI for creating a web server and Poetry for dependency management.

Podman installation

Ubuntu/WSL (using apt)

sudo apt install podman

MacOS (using brew)

brew install podman

Podman configuration

The following commands initialise and run podman as a daemon on your machine.

1. Initialise podman

podman machine init

2. Start podman daemon

podman machine start

Example output:

INFO[0000] waiting for clients...
INFO[0000] listening tcp://127.0.0.1:7777
INFO[0000] new connection from  to /var/folders/n1/wk2myqkx66vf483r8qg7l2880000gn/T/podman/qemu_podman-machine-default.sock
Waiting for VM ...
Machine "podman-machine-default" started successfully

Podman usage

The following podman commands are direct replacements of the Docker CLI. You can see that their syntax is identical:

1. Build image and tag it as fastapi-podman

The following command uses the Dockerfile that is present on the root of the project to build an image.

Notice that there is no difference in the Dockerfile syntax and it can be used 'as is' between Docker and Podman. It still uses Docker Hub to fetch the official Python image that is defined in it (FROM python:3.11).

podman image build -t fastapi-podman .

2. Run a container of the previously tagged image (fastapi-podman)

Run our FastAPI application and map our local port 8000 to 80 on the running container:

podman container run -d --name fastapi-podman -p 8000:80 --network bridge fastapi-podman

3. Check running containers

podman ps

CONTAINER ID  IMAGE                            COMMAND               CREATED         STATUS             PORTS                 NAMES
78586e5b4683  localhost/fastapi-podman:latest  uvicorn main:app ...  13 minutes ago  Up 13 minutes ago  0.0.0.0:8000->80/tcp  nifty_roentgen

4. Hit sample endpoint

Our FastAPI server now runs on port 8000 on our local machine. We can test it with:

curl http://localhost:8000

Output:

{"Hello":"World"}

Integration with AWS

Similarly, the syntax of the commands that are used to push an image to ECR is identical, just replace docker with podman.

Login To ECR

aws ecr get-login-password --region eu-west-2 | podman login -u YOUR_AWS_USERNAME --password-stdin YOUR_ACCOUNT_NUMBER.dkr.ecr.eu-west-2.amazonaws.com

Push to ECR repo

podman push YOUR_ACCOUNT_NUMBER.dkr.ecr.eu-west-2.amazonaws.com/fastapi-podman