Change directory command in Docker?

Mastering Directory Changes in Docker: A Comprehensive Guide

Learn how to effectively manage working directories within Docker containers and Dockerfiles using WORKDIR and other commands, ensuring your applications run smoothly.

When working with Docker, understanding how to navigate and set the working directory within your containers is crucial for building efficient and reliable images. Unlike traditional shell environments where you might frequently use cd, Docker has its own mechanisms for managing the current directory. This article will explore the primary methods for changing directories in Docker, focusing on WORKDIR in Dockerfiles and runtime considerations.

The WORKDIR Instruction in Dockerfiles

The most common and recommended way to set the working directory inside a Docker container is by using the WORKDIR instruction in your Dockerfile. This instruction sets the working directory for any RUN, CMD, ENTRYPOINT, COPY, and ADD instructions that follow it in the Dockerfile. If the WORKDIR does not exist, it will be created automatically.

FROM alpine:latest

# Set the working directory to /app
WORKDIR /app

# Subsequent commands will execute in /app
COPY . .
RUN ls -la

Using WORKDIR in a Dockerfile

How WORKDIR Affects Subsequent Commands

Once WORKDIR is set, all subsequent instructions that operate on the filesystem will use this directory as their base. This simplifies your Dockerfile by allowing you to use relative paths for files and commands. You can use multiple WORKDIR instructions; each one will change the directory for the following instructions relative to the previous WORKDIR if a relative path is provided, or set a new absolute path.

FROM alpine:latest

WORKDIR /app
COPY src/ ./

WORKDIR subfolder
RUN touch my_file.txt # This creates /app/subfolder/my_file.txt

Multiple WORKDIR instructions and their effect

flowchart TD
    A[Dockerfile Start] --> B{WORKDIR /app}
    B --> C[COPY src/ .]
    C --> D{WORKDIR subfolder}
    D --> E[RUN touch my_file.txt]
    E --> F[Result: /app/subfolder/my_file.txt]
    F --> G[Container Runtime]

Flow of WORKDIR instructions in a Dockerfile

Changing Directory at Container Runtime

While WORKDIR is for building the image, you might also need to specify the working directory when running a container. This can be achieved using the -w or --workdir flag with the docker run command. This flag overrides any WORKDIR specified in the Dockerfile for the specific docker run command.

docker run -it --workdir /tmp my_image bash
# Now you are in /tmp inside the container

Setting working directory at container runtime

Why Not Use RUN cd ...?

A common misconception for newcomers is to use RUN cd /path/to/dir in a Dockerfile. This approach is ineffective because each RUN instruction executes in a new shell, and the cd command only changes the directory for that specific shell. The change is lost once that RUN instruction completes. Subsequent RUN instructions will start from the directory set by the last WORKDIR or the default root directory.

FROM alpine:latest

RUN mkdir /mydata
RUN cd /mydata
RUN pwd # This will output '/', not '/mydata'

Incorrect use of RUN cd in a Dockerfile

sequenceDiagram
    participant Dockerfile
    participant Shell1
    participant Shell2

    Dockerfile->>Shell1: RUN mkdir /mydata
    Shell1-->>Dockerfile: Directory created
    Dockerfile->>Shell1: RUN cd /mydata
    Shell1-->>Dockerfile: Shell1's CWD is /mydata
    Note over Shell1: Shell1 exits, CWD reset
    Dockerfile->>Shell2: RUN pwd
    Shell2-->>Dockerfile: Outputs '/'
    Note over Dockerfile: WORKDIR is the correct approach

Why RUN cd is ineffective in Dockerfiles