mikeconrad/demystifying-docker
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Merge branch 'master' into dev-container

Browse Source
This commit is contained in:
Mike Conrad
2025-06-09 16:15:33 -04:00
parent a406b8b11d 1b949309f5
commit 21f8f61f45
1 changed files with 221 additions and 97 deletions
Download Patch File Download Diff File Expand all files Collapse all files

318
slides.md
View File

@ -51,17 +51,62 @@ transition: fade-out
layout: center layout: center
--- ---
## Why Containers? # Who is this for?
- "It works on my machine" is a thing of the past ## About you
- Containers are lightweight and portable - Some experience with Docker/containers
- Boot in milliseconds - Some experience with BASH
- Ideal for reproducible dev environments - Want to better understand how containers work
--- ---
transition: fade-out transition: fade-out
layout: center layout: center
--- ---
## Follow Along
**Example Repo** - https://git.hackanooga.com/mikeconrad/demystifying-docker
---
transition: fade-out
layout: center
---
<img src="https://miro.medium.com/v2/resize:fit:400/format:webp/1*Ibnwjo9LtUFxRY1MZgOcvg.png"/>
---
transition: fade-out
layout: center
---
## Common Use cases for containers
- Reproducible dev environments
- Testing in CI/CD environments
- Better "Portability" of application code
- Snapshot of application code at specific point in time
---
transition: fade-out
layout: center
---
## How we use containers
- PR builds (Preview Environments).
<br />
<br />
### Allows us to
- Test changes in isolated environments
- Simplify complex dev environment setups
- (frontend/backend services, databases, object storage, etc)
<!--A single VM in the cloud running `Docker Compose` with `Traefik` and `Sablier` allows us to have multiple ephemeral preview environments running at any given time.-->
---
transition: fade-out
layout: center
---
## Containers vs Virtual Machines ## Containers vs Virtual Machines
| Feature | VM | Container | | Feature | VM | Container |
@ -71,7 +116,7 @@ layout: center
| Isolation | Strong | Process-level | | Isolation | Strong | Process-level |
| Portability | Medium | Very High | | Portability | Medium | Very High |
In reality we often use containers and vm's together. Containers run inside of VM's for better security and isolation, especially in cloud and multi tenant environments. In reality we use containers and vm's together. Containers run inside of VM's for better security and isolation, especially in cloud and multi tenant environments.
--- ---
transition: fade-out transition: fade-out
@ -80,17 +125,35 @@ layout: center
## What is Docker? ## What is Docker?
- A tool to build and run containers - Written in GO
- Docker engine runs containers using OS features: - Uses Client/Server model with REST API (`docker cli` and `dockerd`)
- Namespaces - Eco system of tools (Compose, Swarm, etc)
- cgroups - Public Image Registry (Dockerhub)
- Union file systems - Docker client typically runs on same machine as server but doesn't have to
- Uses images layered from base -> app code
--- ---
transition: fade-out transition: fade-out
layout: center layout: center
--- ---
## What is Docker?
- A tool to build and run containers
- Containers are exclusive to Linux
- Docker engine runs containers using Linux features like:
- Namespaces
- cgroups
- Union file systems
- Container runs from an image layered with base image and application code
---
transition: fade-out
layout: center
---
## Common Use Cases
- Reproducible Dev environments (dev containers)
- Preview/PR environments (ephemeral test environments)
- Legacy applications or applications with complex environment setups
--- ---
transition: fade-out transition: fade-out
@ -99,10 +162,11 @@ layout: center
## Docker Architecture ## Docker Architecture
Docker Engine (Server) <-- REST API --> Docker CLI (Client) Docker CLI (Client) <-- REST API --> Docker Engine (Server)
<img src="https://docs.docker.com/get-started/images/docker-architecture.webp" width="700" /> <img src="https://docs.docker.com/get-started/images/docker-architecture.webp" width="700" />
<!-- ![Docker Architecture](https://docs.docker.com/get-started/images/docker-architecture.webp) -->
[https://docs.docker.com/get-started/docker-overview/]
--- ---
transition: fade-out transition: fade-out
@ -113,7 +177,10 @@ layout: center
- **Namespaces**: isolate PID, net, mount, etc. - **Namespaces**: isolate PID, net, mount, etc.
- **cgroups**: control CPU, memory, IO - **cgroups**: control CPU, memory, IO
- **UnionFS**: layered filesystem (AUFS, OverlayFS) - **UnionFS**: layered filesystem (OverlayFS)
<!--
Overlayfs is the default. This allows Docker to use a layered approach. In this example the bottom layer or lowerdir is the "filesytem" from the image. The upperdir is the container filesystem (not persisted by default) and the merged is volume/bind mounts.
-->
![UnionFS diagram](https://docs.docker.com/engine/storage/drivers/images/overlay_constructs.webp) ![UnionFS diagram](https://docs.docker.com/engine/storage/drivers/images/overlay_constructs.webp)
@ -122,6 +189,94 @@ transition: fade-out
layout: center layout: center
--- ---
## Bind/Volume Mounts
- 2 most common storage mechanisms
- Different use cases and security implications
---
transition: fade-out
layout: center
---
## Bind Mounts
- Mounting files/directories from the host machine directly into a container (merged overlayfs layer).
- Processes inside container can modify files on host system.
- Bind mounts are strongly tied to the host
- Best for things like dev containers where you need to mount source code into container and have hot reload, etc.
## Bind Mount Example
```bash
$ docker run --mount type=bind,src=/home/mikeconrad/projects/example/app,dst=/app,ro nginx # ro for ReadOnly
$ docker run --volume /home/mikeconrad/projects/example/app:/app nginx
```
<!-- https://docs.docker.com/engine/storage/bind-mounts/ -->
<!--
These terms are oftentimes used interchangebly and can be confusing but it is important to understand the difference.
You need to be careful when using bind mounts because by default, the processes running in the container will have read/write access to your filesystem.
This could cause some issues if code running inside the container is malicious or is compromised.
It is also possible to mount the files as read-only so that the container has access to read them but not write. This is better for security.
Bind mounts also "overwrite" the container/image filesystem layers. So for example mounting ./some-files/test:/etc/passwd would overwrite the /etc/passwd file in the container
The directory inside the container does not need to exist for this to work. If the directory does not exist inside of the container filesystem it will be created with the contents.
-->
---
transition: fade-out
layout: center
---
## Volume Mount Example
```bash
$ docker run --name postgrestest \
--mount type=volume,src=postgresData,dst=/var/lib/postgresql/data \
-e POSTGRES_PASSWORD=postgres \
--rm postgres:16
$ docker run --name postgrestest \
--volume postgresData:/var/lib/postgresql/data \
-e POSTGRES_PASSWORD=postgres \
--rm postgres:16
```
```bash
$ docker volume inspect postgresData
[
{
"CreatedAt": "2025-06-08T10:39:12-04:00",
"Driver": "local",
"Labels": null,
"Mountpoint": "/var/lib/docker/volumes/postgresData/_data",
"Name": "postgresData",
"Options": null,
"Scope": "local"
}
]
```
- Docker creates a volume named postgresData and mounts that directory inside the container.
<!-- https://docs.docker.com/engine/storage/bind-mounts/ -->
---
transition: fade-out
layout: center
---
## Volume mounts
- Created and managed by the Docker Daemon
- Volume data is stored on host filesystem but managed by Docker.
- Used for persistent data.
<!--
It is possible to modify the data directly via normal tools but unsupported and can cause unintended side-effects due to the overlayfs storage driver.
An example would be creating a postgres volume for persistent database storage.
-->
---
transition: fade-out
layout: center
---
## Anatomy of a Dockerfile ## Anatomy of a Dockerfile
```dockerfile ```dockerfile
@ -133,6 +288,11 @@ COPY . .
EXPOSE 3000 EXPOSE 3000
CMD ["npm", "start"] CMD ["npm", "start"]
``` ```
```bash
mikeconrad@pop-os:~/projects/demystifying-docker/examples/react
$ docker build -t react-app .
```
- Starts with a base image - Starts with a base image
- Copy files and install deps - Copy files and install deps
@ -143,33 +303,35 @@ transition: fade-out
layout: center layout: center
--- ---
## Dockerfile Best Practices ## Multi Stage builds
```dockerfile ```dockerfile
# Stage 1: Build the Go binary # Stage 1 - Define Base image
FROM golang:1.24.2-alpine AS builder FROM node:22-alpine AS base
# Set working directory inside the build container # Stage 2 Install dependencies
FROM base AS install-deps
WORKDIR /app WORKDIR /app
COPY go.mod ./ COPY package*.json /app/
RUN go mod download RUN yarn
# Stage 3 Development
FROM install-deps AS develop
WORKDIR /app
COPY --from=install-deps /app/node_modules /app/node_modules
COPY . . COPY . .
# Build the Go binary statically ENTRYPOINT ["yarn", "dev", "--host=0.0.0.0"]
RUN CGO_ENABLED=0 GOOS=linux go build -o docker-api-proxy . EXPOSE 5173
# Stage 2: Run binary in minimal container
FROM scratch
# Copy binary from builder
COPY --from=builder /app/docker-api-proxy /usr/local/bin/docker-api-proxy
# Run binary
ENTRYPOINT ["docker-api-proxy"]
EXPOSE 80
``` ```
- Use specific versions, not `latest` ```bash
- Combine commands to reduce layers $ docker build -t react .
- Use `.dockerignore` $ docker run --rm -P react
- Prefer slim or alpine images ```
- Run as non-root user if possible <!--
Run docker image and demonstrate dev container functionality. Attach to the running container
via VSCode extension and make changes to code. Note that it updates in real time in the browser.
Kill the container and start a new one. Note that the files do not persist. Need volume/bind mounts
For that.
-->
--- ---
transition: fade-out transition: fade-out
layout: center layout: center
@ -180,64 +342,26 @@ layout: center
- Define multi-container apps in one file - Define multi-container apps in one file
- Great for local dev and staging (and production!) - Great for local dev and staging (and production!)
```yaml ---
name: traefik_secure transition: fade-out
services: layout: center
socket-proxy: ---
image: git.hackanooga.com/mikeconrad/docketproxy:latest
container_name: socket-proxy ## Q/A
networks:
- traefik -
- socket_proxy
volumes: ---
- /var/run/docker.sock:/var/run/docker.sock:ro transition: fade-out
read_only: true layout: center
security_opt: ---
- no-new-privileges:true
cap_drop: ## Resources
- ALL - [Slide Deck (including examples)](https://git.hackanooga.com/mikeconrad/demystifying-docker-v2)
restart: unless-stopped - [DocketProxy (Docker socket proxy)](https://git.hackanooga.com/mikeconrad/docketproxy)
environment: - [SlimToolkit (Optimize and secure containers)](https://github.com/slimtoolkit/slim)
- ALLOWED_NETWORKS=traefik_secure_traefik
traefik: ## VSCode plugins
image: traefik:latest https://marketplace.visualstudio.com/items?itemName=ms-azuretools.vscode-docker
container_name: traefik https://marketplace.visualstudio.com/items?itemName=ms-vscode-remote.remote-containers
command: https://marketplace.visualstudio.com/items?itemName=ms-azuretools.vscode-containers
- "--log.level=INFO"
- "--entrypoints.web.address=:80"
- "--entrypoints.websecure.address=:443"
- "--providers.docker=true"
- "--providers.docker.endpoint=tcp://socket-proxy:8000"
- "--providers.docker.exposedbydefault=false"
- "--entrypoints.traefik.address=:8080"
- "--api.insecure=true"
- "--api.dashboard=true"
labels:
- "traefik.enable=true"
- "traefik.http.routers.api.rule=Host(`traefik.docker.localhost`)"
- "traefik.http.routers.api.entrypoints=web"
- "traefik.http.routers.api.service=api@internal"
ports:
- "80:80"
- "8080:8080"
networks:
- traefik
- socket_proxy
depends_on:
- socket-proxy
restart: unless-stopped
whoami:
image: traefik/whoami
networks:
- traefik
labels:
- "traefik.enable=true"
- "traefik.http.routers.whoami.rule=Host(`whoami.docker.localhost`)"
- "traefik.http.routers.whoami.entrypoints=web"
networks:
traefik: {}
socket_proxy:
driver: bridge
internal: true
enable_ipv6: false
```