Docker Bind Mounts: Share Your Code with a Container

A bind mount connects a folder on your computer to a path in a container. Both see the same files. This is ideal for development, because you can edit code on your computer, and the container sees the change straight away.

Syntax#

If the left side of -v is a path, it is a bind mount. If it is a name, it is a volume:

docker run -v "$PWD/src":/app/src image     # bind mount
docker run -v mydata:/app/data image        # named volume

Live changes#

Here Nginx serves a folder from the host, with no custom image:

mkdir html
echo "<h1>Version 1</h1>" > html/index.html
docker run -d --name web -p 8080:80 -v "$PWD/html":/usr/share/nginx/html:ro nginx:1.30-alpine > /dev/null
sleep 2
curl -s http://localhost:8080
echo "<h1>Version 2</h1>" > html/index.html
curl -s http://localhost:8080
docker rm -f web > /dev/null

<h1>Version 1</h1>
<h1>Version 2</h1>

The page changed with no rebuild and no restart. :ro stops the container from changing your files.

Using tools without installing them#

A bind mount lets you run tools on your files from a container:

mkdir work
echo "Hello from the host" > work/in.txt
docker run --rm -v "$PWD/work":/work -w /work alpine:3.24 sh -c 'tr a-z A-Z < in.txt > out.txt'
cat work/out.txt

HELLO FROM THE HOST

-w sets the working folder. This is a handy way to use a specific Node.js or Python version without installing it.

File ownership#

On Linux, files created by a container are owned by the container's user, which is often root:

docker run --rm -v "$PWD/work":/work alpine:3.24 touch /work/made-by-root
ls -l work/made-by-root | awk '{print $3, $4}'
docker run --rm --user "$(id -u):$(id -g)" -v "$PWD/work":/work alpine:3.24 touch /work/made-by-me
ls -l work/made-by-me | awk '{print $3, $4}'

root root
asha asha

--user runs the container with your user ID, which avoids files that you cannot delete.

The --mount syntax#

--mount is longer, but clearer. Unlike -v, it fails if the host path does not exist, rather than creating an empty folder:

docker run --rm --mount type=bind,src="$PWD/work",dst=/work,readonly alpine:3.24 ls /work | head -n 1
docker run --rm --mount type=bind,src=/no/such/path,dst=/work alpine:3.24 true 2>&1 | head -n 1 | cut -c1-96

in.txt
docker: Error response from daemon: invalid mount config for type "bind": bind source path does

tmpfs#

A tmpfs mount lives in memory, and vanishes when the container stops. It suits temporary files and secrets that should never reach the disk:

docker run --rm --tmpfs /scratch alpine:3.24 sh -c 'mount | grep scratch | cut -d, -f1'

tmpfs on /scratch type tmpfs (rw

Which should I use?#

A mount hides what was there#

A mount covers whatever the image had at that path. If you mount your project over /app, the node_modules installed in the image is hidden. Compose for Development shows the fix.

On Windows and macOS#

Bind mounts cross from your computer into Docker's Linux virtual machine, so they can be slower, and file-change events may not arrive. On Windows, keep the project inside the WSL file system, not on /mnt/c.

Common mistakes#

Practice#

mkdir conf
echo "mode=test" > conf/app.txt
docker run --rm -v "$PWD/conf":/conf:ro alpine:3.24 sh -c 'cat /conf/app.txt; touch /conf/x'

mode=test
touch: /conf/x: Read-only file system

Key takeaway#

Next: Container Networking.