How do I prevent docker:exec_not_found in CI?

Add a simple smoke test in CI that runs the image and checks the entrypoint. Prefer absolute paths and ensure scripts have correct line endings and shebang.

exec: "python": executable file not found in $PATH

Q: How can I manually validate docker:exec_not_found?

Inspect the image entrypoint and command with `docker image inspect ` so you know what Docker is trying to execute. Start a shell in the image, if possible, and verify the target path exists, has execute permissions, and uses the expected shebang or binary format. Check whether a bind mount or copied script replaced the original executable with a file that has different permissions or line endings.

Fix it fast

Most likely: The command in ENTRYPOINT, CMD, docker run, or docker exec is not installed in the image, is not on PATH, or needs a missing interpreter such as python or sh.

1. Confirm this is your error

exec: "python": executable file not found in $PATH

2. Check the cause

docker image inspect <image>
docker run --rm --entrypoint sh <image> -lc 'echo "$PATH"; command -v <binary>; ls -la <path>'
docker run --rm --entrypoint sh <image> -lc 'head -1 <script-path>'

3. Apply the safe fix

# Install the missing binary or interpreter in the final image stage.
# Or change ENTRYPOINT/CMD to the exact path that exists in the image.

# Example for Python images: use python3 if the image does not provide python.
docker build -t <image> .
docker run --rm <image>

4. Verify it works

docker run --rm --entrypoint sh <image> -lc 'command -v <binary>'
docker run --rm <image>

Don't use unsafe shortcuts

Do not assume commands installed on your host exist inside the container image.
Do not debug registry or Docker daemon settings for this error, the container is already starting.
Do not forget multi-stage builds: the binary may exist in the build stage but not in the final image.

What Docker Could Not Execute

Docker started the container, but the configured command or entrypoint was not found in the image PATH or at the specified path.

Fix the container entrypoint or command

Inspect the Dockerfile ENTRYPOINT/CMD and confirm the file exists in the final image, enter the image and check with docker run --rm -it <image> sh then ls -la <path>, and fix permissions in the image build (example with RUN chmod +x /path/to/file).

Manual container debug checklist

Inspect the image entrypoint and command with docker image inspect <image> so you know what Docker is trying to execute.

Start a shell in the image, if possible, and verify the target path exists, has execute permissions, and uses the expected shebang or binary format.

Check whether a bind mount or copied script replaced the original executable with a file that has different permissions or line endings.

Why the Container Could Not Start

Usually this comes down to the binary/script is not present in the image at the expected path, file permissions do not allow execution (chmod +x missing), or the command is correct but depends on a missing runtime/interpreter.

Verify the Container Starts Cleanly

Rebuild the image and re-run the container, and confirm the container process starts and stays running.

How Docker starts containers

This failure happens after Docker has already selected the image and is trying to start the process inside the container. At that point the important questions are whether the entrypoint or command exists at the expected path, whether it is executable, and whether the container filesystem or mounted volume changed what Docker is able to run.

Prevent Container Startup Regressions

To prevent this, add a simple smoke test in CI that runs the image and checks the entrypoint, and prefer absolute paths and ensure scripts have correct line endings and shebang.