I have following Dockerfile:
FROM ubuntu:16.04
ARG path1=def_path1
RUN mkdir ${path1}
When I build this Dockerfile using following command:
docker build --build-arg path1=/home/dragan -t build_arg_ex .
I get following error when I execute it in MINGW bash on Windows 10:
$ ./build.sh --no-cache
Sending build context to Docker daemon 6.144kB
Step 1/3 : FROM ubuntu:16.04
---> 2a4cca5ac898
Step 2/3 : ARG path1=def_path1
---> Running in a35241ebdef3
Removing intermediate container a35241ebdef3
---> 01475e50af4c
Step 3/3 : RUN mkdir ${path1}
---> Running in 2759e683cbb1
mkdir: cannot create directory 'C:/Program': No such file or directory
mkdir: cannot create directory 'Files/Git/home/dragan': No such file or
directory
The command '/bin/sh -c mkdir ${path1}' returned a non-zero code: 1
Building same Dockerfile in Windows Command Prompt or on Linux or Mac is ok. The problem is only in MINGW bash terminal on Windows because it adds 'C:/Program Files/Git' before the path that is passed as argument.
Is there a way to execute this in MINGW bash so it does not add the 'C:/Program Files/Git' prefix?
Thanks
This is actually a bug/limitation of Git for Windows as described in the Release Notes under Known issues:
If you specify command-line options starting with a slash, POSIX-to-Windows path conversion will kick in converting e.g. "/usr/bin/bash.exe" to "C:\Program Files\Git\usr\bin\bash.exe". When that is not desired -- e.g. "--upload-pack=/opt/git/bin/git-upload-pack" or "-L/regex/" -- you need to set the environment variable MSYS_NO_PATHCONV temporarily, like so:
MSYS_NO_PATHCONV=1 git blame -L/pathconv/ msys2_path_conv.cc
Alternatively, you can double the first slash to avoid POSIX-to-Windows path conversion, e.g. "//usr/bin/bash.exe".
Further to #mat007's answer:
This bash function solved the problem more permanently for docker, without enabling MSYS_NO_PATHCONV globally, which causes another world of pain.
.bashrc
# See https://github.com/docker/toolbox/issues/673#issuecomment-355275054
# Workaround for Docker for Windows in Git Bash.
docker()
{
(export MSYS_NO_PATHCONV=1; "docker.exe" "$#")
}
You may need to do the same for docker-compose
Related
Following the instructions as outlined to deploy Duo CloudMapper to AWS environment and getting an error
Docker File
FROM python:3.7-slim as cloudmapper
LABEL maintainer="https://github.com/0xdabbad00/"
LABEL Project="https://github.com/duo-labs/cloudmapper"
WORKDIR /opt/cloudmapper
ENV AWS_DEFAULT_REGION=us-east-1
RUN apt-get update -y
RUN apt-get install -y build-essential autoconf automake libtool python3.7-dev python3-tk jq awscli
COPY cloudmapper/. /opt/cloudmapper
COPY entrypoint.sh /opt/cloudmapper/entrypoint.sh
# Remove the demo data
RUN rm -rf /opt/cloudmapper/account-data/demo
# Install the python libraries needed for CloudMapper
RUN cd /opt/cloudmapper && pip install -r requirements.txt
ENTRYPOINT /opt/cloudmapper/entrypoint.sh
Now building the docker image
C:\> docker build -t cloudmapper .
When I run the docker using the below command I get an error
C:/> docker run -t cloudmapper
Error
/bin/sh: 1: /opt/cloudmapper/entrypoint.sh: not found
Verified that the file exists in the appropriate location
Using Docker on Windows 10
Image in the dockerfile is python:3.7-slim
Assuming the images are removed and replaced with text and the question doesn't get closed.
bash can return "file not found" when
the entrypoint shell script is not marked executable for the current user
the hash bang in the entrypoint shell script points to a binary that does not exist
the shell script actually does not exist.
You can fix the first problem by ensuring you use the new --chmod flag to ensure the executable bit is set. Even if the user is root it is necessary that there is at least 1 executable bit set.
COPY --chmod=0755 *.sh /opt/cloudmapper/
ENTRYPOINT ["/opt/cloudmapper/entrypoint.sh"]
ps. This integrated COPY --chmod only works with buildkit enabled builds, so you might need to force buildkit, or split the chmod into a separate explicit RUN step.
The 2nd issue can be dealt with by ensuring the first line of entrypoint.sh uses sh rather than bash if you are using a lightweight base image like alpine:
#!/bin/sh
set -e
# etc
Also, if on Windows especially, ensure ALL files, especially the entrypoint .sh file, are set to utf-8 encoding with lf style line endings. As linux doesn't understand the cr, it will try to execute /bin/sh<cr> as the shell which clearly doesn't exist.
In terms of the file not existing, verify the entrypoint.sh is being copied into a location that is referenced by env.PATH, or that the entry point directive uses a fully qualified path.
--
edited to add cr-lf revelation.
I have bash script that performing some Docker commands:
#!/usr/bin/env bash
echo "Create and start database"
cd ../../database
cp -R ../../../scripts/db db/
docker build -t a_database:1 .
docker run --rm --name a_db -e POSTGRES_PASSWORD=docker -d -p 5432:5432 a_database:1
docker network connect --ip 172.23.0.5 a_network a_db
sleep 15
echo "Initialize database"
docker exec a_db /root/db/dev/init_db.sh
echo "Cleanup"
rm -rf db
On mac everything works fine, problem occurs when I try to start this script on windows machine. When I'm running it I receive an error:
OCI runtime exec failed: exec failed: container_linux.go:344: starting container process caused "exec: \"C:/Program Files/Git/root/db/dev/init_db.sh\": stat C:/Program Files/Git/root/db/dev/init_db.sh: no such file or directory": unknown
Directory and script (/root/db/dev/init_db.sh) exist inside docker container. I don't know why it tries to find script on host machine? Also when I perform command:
docker exec a_db /root/db/dev/init_db.sh
directly in command line (on windows) script is executed. Any idea what is wrong and why it's trying to use git ?
I had a similar problem... absolute paths with windows variables fixed mine:
$HOME/docker/...
Thanks to igaul answer I was able to run this on windows machine. There were two problems:
Path to script in docker container. Instead of:
docker exec a_db /root/db/dev/init_db.sh
should be:
docker exec a_db root/db/dev/init_db.sh
Line endings in init_db.sh. On windows machine after pulling repository from bitbucket line ending of init_db.sh was setup to CRLF what caused problem. I've added .gitattribute file to my repo and now init_db.sh file always has LF endings.
It's not a bug in Docker, but the way mingw handles these paths. Here is some more information about that "feature"; http://www.mingw.org/wiki/Posix_path_conversion. Prefixing the path with a double slash (//bin/bash) should prevent this, or you can set MSYS_NO_PATHCONV=1, see How to stop MinGW and MSYS from mangling path names given at the command line
[this is the error I'm getting after build command ]
Step 7/9 : RUN chmod +x /main.sh
---> Running in 6e880a009c7d
chmod: cannot access '/main.sh': No such file or directory
The command '/bin/sh -c chmod +x /main.sh' returned a non-zero code: 1
and here is my docker file
FROM centos:latest
MAINTAINER Aditya Gupta
#install git
RUN yum -y update
RUN yum -y install git
#make git repo folder, change GIT_LOCATION
RUN mkdir -p /home/centos/doimages/dockimg;cd /home/centos/doimages/dockimg;
RUN git clone https://(username):(password)#gitlab.com/abc/xyz.git (foldername);cd (foldername)/
Run chmod +x ./main.sh
RUN echo " ./main.sh\n "
EXPOSE Portnumber
When you perform a RUN step in a Dockerfile, a temporary container is launched, often with a shell parsing your command. When that command finishes, the container exits, and docker packages the filesystem changes as an image layer. That process is repeated from the beginning for each RUN line.
The key piece there is the shell exits, losing environment variables you've set, background processes you've run, and in this case, the current working directory you tried to set here:
RUN git clone https://(username):(password)#gitlab.com/abc/xyz.git (foldername);cd (foldername)/
Instead of a cd in a RUN command, you can update the value of WORKDIR:
RUN git clone https://(username):(password)#gitlab.com/abc/xyz.git (foldername)
WORKDIR foldername
You want to execute a shell file which does not exist on your docker machine. use ADD command to add your script to your docker image!
-- somewehe inside your dockerfile befor the execution ---
ADD ./PATH/ON/HOST/main.sh /PATH/YOU/LIKE/ON/DOCKER/MACHINE
Then try to build your docker machine
issue is resolved with workdir and cloning manually without docker file and then give the path to mainsh in dockerfile.
So I am running a script that calls:
docker-compose run --rm web sh /data/bin/install_test_db.sh
and it seems to work fine when running it on ubuntu but when I run it on Windows using Git Bash I get this error:
sh: 0: Can't open C:/Program Files/Git/data/bin/install_test_db.sh
It seems to be trying to run the script on my machine and not the actual container from which I am trying to call it from.
Any ideas on why this is happening?
First, try passing a command to your sh shell:
docker-compose run --rm web sh -c "/data/bin/install_test_db.sh"
Or:
docker-compose run --rm web "sh -c /data/bin/install_test_db.sh"
That will avoid your host shell (the git bash) to interpret an absolute path (starting with '/') as one from the Git installation path.
That will keep /data/... as an argument to be passed to the shell executed in the container.
Note: If you are using Docker for Windows (meaning not VirtualBox, but HyperV), you don't need git bash at all.
Try the same command from a regular CMD (with docker.exe in your %PATH%, which Docker for Windows set for you)
vonc#VONCAVN7 C:\Users\vonc
> where docker
C:\Program Files\Docker\Docker\Resources\bin\docker.exe
If you need Linux-like commands from that same CMD session, then yes, add git paths:
set GH=C:\path\to\git
set PATH=%GH%\bin;%GH%\usr\bin;%GH%\mingw64\bin;%PATH%
Running Docker Toolbox on Windows 10 host.
There is a Dockerfile:
FROM 16.04
...
RUN if [ some_condition ]; then ./foo.sh; fi
...
There is a foo.sh:
#!/bin/bash
...
echo 'Me working'
Now when trying to build the Docker image:
docker build -t name_of_the_image .
Getting error:
Step 7/12 : RUN ./foo.sh
---> Running in e7e0703d3f8f
/bin/sh: 1: ./foo.sh: not found
The command '/bin/sh -c ./foo.sh' returned a non-zero code: 127
I would assume error 127 would be the Docker doesn't see the bash. Any suggestion how to fix this?
Edit: already copying all files into the Docker, Dockerfile:
FROM ubuntu:16.04
MAINTAINER Mr Anderson "mr#anderson.com"
# set workdir
COPY . /app
WORKDIR /app
# Run scripts
RUN ./foo.sh
You'll need to copy/COPY the file into the container before you can execute/RUN the script.
Also since you're using a relative path when you call the script be sure to set a WORKDIR.
COPY ./foo.sh /app/foo.sh
WORKDIR /dir
RUN chmod +x /app/foo.sh
RUN if [ some_condition ]; then ./foo.sh; fi
Also make sure the script is executable.
After some further investigation:
Using CMD over RUN is not a perfect solution because of the way those commands work. RUN can be used any amount of times, to build Docker image layer by layer, while CMD can be executed only once when the image has been build.
In my case the solution was to:
Open ./foo.sh file with VIM and run: :set fileformat=unix and save the file.
Long story short: the line ending in the shell-script were incorrect and had to be converted to the Unix ones.