docker-compose "No such file or directory" for sh-command

Question

I am setting up docker-for-windows on my private pc.

When I set it up a while ago on my office laptop I had the same issue but it just stopped happening.

So I am stuck with this:

I have a docker-working project (on my other computer) with a docker-compose.yml like this:

version: '2'

services:
  web:
    depends_on:
      - db
    build: .
    env_file: ./docker-compose.env
    command: bash ./run_web_local.sh
    volumes:
      - .:/srv/project
    ports:
      - 8001:8001
    links:
      - db
      - rabbit
    restart: always

Dockerfile:

### STAGE 1: Build ###

# We label our stage as 'builder'
FROM node:8-alpine as builder

RUN npm set progress=false && npm config set depth 0 && npm cache clean --force

# build backend
ADD package.json /tmp/package.json
ADD package-lock.json /tmp/package-lock.json
RUN cd /tmp && npm install
RUN mkdir -p /backend-app && cp -a /tmp/node_modules /backend-app

### STAGE 2: Setup ###
FROM python:3

# Install Python dependencies
COPY requirements.txt /tmp/requirements.txt
RUN pip3 install -U pip
RUN pip3 install --no-cache-dir -r /tmp/requirements.txt

# Set env variables used in this Dockerfile (add a unique prefix, such as DOCKYARD)
# Local directory with project source
ENV PROJECT_SRC=.
# Directory in container for all project files
ENV PROJECT_SRVHOME=/srv
# Directory in container for project source files
ENV PROJECT_SRVPROJ=/srv/project

# Create application subdirectories
WORKDIR $PROJECT_SRVPROJ
RUN mkdir media static staticfiles logs

# make folders available for other containers
VOLUME ["$PROJECT_SRVHOME/media/", "$PROJECT_SRVHOME/logs/"]

# Copy application source code to SRCDIR
COPY $PROJECT_SRC $PROJECT_SRVPROJ

COPY --from=builder /backend-app/node_modules $PROJECT_SRVPROJ/node_modules

# Copy entrypoint script into the image
WORKDIR $PROJECT_SRVPROJ

# EXPOSE port 8000 to allow communication to/from server
EXPOSE 8000

CMD ["./run_web.sh"]

docker-compose.env:

C_FORCE_ROOT=True
DJANGO_CELERY_BROKER_URL=amqp://admin:mypass@rabbit:5672//
DJANGO_DATABASE_ENGINE=django.db.backends.mysql
DJANGO_DATABASE_NAME=project-db
DJANGO_DATABASE_USER=project-user
DJANGO_DATABASE_PASSWORD=mypassword
DJANGO_DATABASE_HOST=db
DJANGO_ALLOWED_HOSTS=127.0.0.1,localhost
DJANGO_DEBUG=True
DJANGO_USE_DEBUG_TOOLBAR=off
DJANGO_TEST_RUN=off
PYTHONUNBUFFERED=0

run_web_local.sh:

#!/bin/bash
echo django shell commands
python ./manage.py migrate
echo Starting django server on 127.0.0.1:8000
python ./manage.py runserver 127.0.0.1:8000

When I call docker-compose up web I get the following error:

web_1 | bash: ./run_web_local.sh: No such file or directory

• I checked the line endings, they are UNIX
• the file exists on the file system as well as inside the container
• I can call bash run_web_local.sh from my windows powershell and inside the container
• I changed the UNIX permissions inside the container
• I left out the bash in the command in the docker-compose command. And tried with backslash, no dot etc.
• I reinstalled docker
• I tried switching to version 3
• docker claims to have a connection to my shared drive C

And: The exact same setup works on my other laptop.

Any ideas? All the two million github posts didn't solve the problem for me.

Thanks!

Update

Removing volumes: from the docker-compose makes it work like stated here but I don't have an instant mapping. That's kind of important for me...

Answer 1

I had same issue recently and the problems goes away using any advanced editor and changing line ending to unix style on sh entrypoint scripts.

In my case, not sure why, because git handle it very well depending on linux or windows host I ended up in same situation. If you have files mounted in container and host(in windows) it dependes on how you edit them may be changed to linux from inside container but that affects outside windows host. After that git stated file .sh changed but no additions no deletions. a graphical compare tool showed up that only new line where changed.

Another workaround to trouble shooting you can start your container overriden entrypoint script by sh for example and then from there you can check on the started container how linux sees the entrypoint script, even you can test it and will see exact same error

Sorry for my english, not my native language. Hope this helps someone.