Scratch base image for minimal containers in Docker - Step-by-Step Execution

Process Flow - Scratch base image for minimal containers

Start Dockerfile

↓

Use FROM scratch

↓

Add minimal files

↓

Build container image

↓

Run minimal container

↓

Container runs with minimal size

↓

End

This flow shows how a Dockerfile starts with FROM scratch, adds only needed files, builds the image, and runs a minimal container.

Execution Sample

Docker

FROM scratch
ADD hello /
CMD ["/hello"]

This Dockerfile uses scratch as base, adds a hello binary, and runs it.

Process Table

Step	Dockerfile Instruction	Action	Resulting Image State	Notes
1	FROM scratch	Start from empty base image	Image has no files or layers	Base image is empty, no OS or tools
2	ADD hello /	Add hello binary to root	Image now contains /hello file	Only the hello binary is added, minimal size
3	CMD ["/hello"]	Set default command to run /hello	Container will run /hello on start	No shell or other commands available
4	docker build -t hello-minimal .	Build image from Dockerfile	Image built with minimal size	Image size is very small, only hello binary
5	docker run --rm hello-minimal	Run container from image	Container runs and executes /hello	Container runs with minimal environment
6	Exit	Container stops after /hello finishes	Container exits cleanly	No extra processes or files remain

💡 Container exits after running the /hello binary, no OS or shell present

Status Tracker

Variable	Start	After Step 2	After Step 3	After Build	After Run	Final
Image Files	empty	/hello	/hello	/hello	/hello (running)	empty (container exited)
Container State	none	none	none	built	running /hello	exited

Key Moments - 3 Insights

Why does the image have no OS or shell?

Because FROM scratch starts with an empty image, no OS or shell is included, only what you add explicitly (see execution_table step 1).

How can the container run a command without a shell?

The CMD runs the binary directly; no shell is needed because the binary is self-contained (see execution_table step 3).

Why is the image size so small?

Only the hello binary is added, no extra OS files or libraries, so the image stays minimal (see execution_table step 2 and 4).

Visual Quiz - 3 Questions

Look at the execution table, what files exist in the image after step 2?

A Only /hello binary

B Full OS files

C No files at all

D Shell and utilities

💡 Hint

Check the 'Resulting Image State' column at step 2 in the execution_table

At which step does the container start running the /hello binary?

A Step 3

B Step 4

C Step 5

D Step 2

💡 Hint

Look at the 'Action' and 'Resulting Image State' columns in execution_table step 5

If you add a shell binary in step 2, how would the image size change?

A It would stay the same

B It would become larger

C It would become smaller

D It would fail to build

💡 Hint

Adding more files increases image size, see step 2 notes about minimal size

Concept Snapshot

FROM scratch starts an empty image with no OS.
ADD copies only needed files (like a binary).
CMD sets the command to run inside the container.
Result: minimal container size, no shell or OS included.
Use for very small, secure containers.

Full Transcript

This visual execution shows how to create a minimal Docker container using the scratch base image. The Dockerfile starts with FROM scratch, which means the image is empty with no operating system or shell. Then, the ADD instruction copies a single binary file named hello into the image. The CMD instruction sets the container to run this hello binary when started. Building the image creates a very small container image containing only the hello binary. Running the container executes the binary directly, without any shell or extra OS files. The container exits after the binary finishes. This approach is useful for creating minimal, secure containers with only the necessary files.