Error in building docker image with perl installation - docker

I am new to docker and i am trying to build the docker image with the perl installation but not sure exactly how to fix this error.
Dockerfile:
FROM amazonlinux
RUN mkdir /shared
RUN cd /shared
RUN yum -y install sudo
RUN cd /shared
RUN echo "Installing Perl."
RUN sudo yum -y update; yum -y install gcc
RUN yum -y install wget
RUN wget http://www.cpan.org/src/5.0/perl-5.22.1.tar.gz
RUN tar -xzf perl-5.22.1.tar.gz
RUN cd perl-5.22.1
RUN /shared/perl-5.22.1/Configure -des -Dprefix=/opt/perl-5.22.1/localperl
RUN make
RUN make test
RUN make install
RUN echo "Perl installation complete."
instead of /shared/perl-5.22.1/Configure i tried to give ./configure as well but i get the same error No such file or directory
Error:
/bin/sh: /shared/perl-5.22.1/Configure: No such file or directory
The command '/bin/sh -c /shared/perl-5.22.1/Configure -des -Dprefix=/opt/perl-5.22.1/localperl' returned a non-zero code: 127
ish-mac:testanalyse ish$
Can anyone tell me how to fix this issue.

Each Dockerfile RUN command runs in its own shell. So, when you do something like RUN cd /shared, the subsequent RUN commands will not be run inside that working directory.
What you want to use in this case is the WORKDIR instruction (https://docs.docker.com/engine/reference/builder/#workdir). You can also combine and shorten some things by taking advantage of the ADD instruction. A more concise Dockerfile to do what you are after might be:
FROM amazonlinux
WORKDIR /shared
RUN yum -y install gcc
ADD http://www.cpan.org/src/5.0/perl-5.22.1.tar.gz /shared
RUN tar -xzf perl-5.22.1.tar.gz
RUN /shared/perl-5.22.1/Configure -des -Dprefix=/opt/perl-5.22.1/localperl
RUN make -C /shared/perl-5.22.1
RUN make -C /shared/perl-5.22.1 test
RUN make -C /shared/perl-5.22.1 install
For example, the build context is already running as root so there is no need for sudo. With ADD we can add directly from URLs and no wget is required. And the make utility has a -C option to specify the working directory for make.
This should get you closer to what you are after. But the build still fails for other reasons (which you should probably open another question for if you are stuck).

Related

qemu-x86_64: Could not open '/lib64/ld-linux-x86-64.so.2': No such file or directory

I have a Rancher Deskop(dockerd) on M1 MacOS and when I am trying to build below dockerfile I am getting an error such as below. Here is the command how I am trying to build the image docker build -t te-grafana-dashboards-toolchain --no-cache .
I tried to change the platforms but nonae of them worked for me. I am a bit lost about this platform issue for M1 but any help will be appreciated, What I am doing wrong? What might the root cause of this?
Removing intermediate container 70af516d5d6b
---> a69229847153
Step 5/6 : RUN GO111MODULE="on" go get github.com/jsonnet-bundler/jsonnet-bundler/cmd/jb; ln -s $(go env GOPATH)/bin/jb /usr/bin/jb
---> Running in 13545862fffe
qemu-x86_64: Could not open '/lib64/ld-linux-x86-64.so.2': No such file or directory
qemu-x86_64: Could not open '/lib64/ld-linux-x86-64.so.2': No such file or directory
Removing intermediate container 13545862fffe
Dockerfile
FROM --platform=linux/amd64 ubuntu:focal
RUN apt update; apt install -y curl jq build-essential python3.8 python3-pip docker-compose jsonnet bison mercurial
RUN ln -s /usr/bin/python3.8 /usr/bin/python
RUN curl -OL https://golang.org/dl/go1.17.linux-amd64.tar.gz; mkdir /etc/golang; tar -xvzf go1.17.linux-amd64.tar.gz -C /etc/golang; ln -s /etc/golang/go/bin/go /usr/bin/go; rm -f go1.17.linux-amd64.tar.gz
RUN GO111MODULE="on" go get github.com/jsonnet-bundler/jsonnet-bundler/cmd/jb; ln -s $(go env GOPATH)/bin/jb /usr/bin/jb
WORKDIR /workspace
Incidentally, in case it's helpful to another who lands here, I have the same issue on an M1 Max MacBook Pro laptop attempting to do a docker build from a company repo that should be a pretty well traveled path, but I might be the only one (it's a small company) that has an ARM64 M1 "Apple Silicon" Mac. However I found the solution (well, a solution) to my situation was exactly the opposite of the solution to the OP's, and that was to add --platform=linux/amd64 to the FROM line of the docker file.
Otherwise it was using an ARM64 image to start from without me being the wiser but then later in the Dockerfile the build attempts to install and execute code compiled for x86_64. Starting the build process by requesting the base image be linux/amd64 ends up with then the base image having /lib64/ld-linux-x86-64.so.2. This probably means everything is being emulated as x86_64 on the ARM64 CPU via qemu-x86_64 and so if you have the option to start from an ARM64 image and can compile within the container during build time any software you can't install as ARM64 binaries, it'll probably go faster when you later run the container on the M1 based Mac. I'm not able to try that myself just yet for this case.
Modifying Dockerfile seems to be the most popular answer but you can also set the DOCKER_DEFAULT_PLATFORM environment variable to linux/amd64.
export DOCKER_DEFAULT_PLATFORM=linux/amd64
The cause seems to reside in the AArch64 image.
this resolved my issue.
FROM ubuntu:focal
RUN apt update; apt install -y curl jq build-essential python3.8 python3-pip docker-compose jsonnet bison mercurial
RUN ln -s /usr/bin/python3.8 /usr/bin/python
RUN curl -OL https://golang.org/dl/go1.17.linux-arm64.tar.gz; mkdir /etc/golang; tar -xvzf go1.17.linux-arm64.tar.gz -C /etc/golang; ln -s /etc/golang/go/bin/go /usr/bin/go; rm -f go1.17.linux-arm64.tar.gz
RUN GO111MODULE="on" go get github.com/jsonnet-bundler/jsonnet-bundler/cmd/jb#latest; ln -s /root/go/bin/jb /usr/bin/jb
WORKDIR /workspace
Instead of editing the Dockerfile, as suggested in this answer, or setting an environment variable, as suggested in this answer, I prefer to pass the platform as an argument to the docker build command, with the --platform flag. The command used by the OP would then be:
docker build --platform linux/amd64 -t te-grafana-dashboards-toolchain --no-cache .
Passing following flag to C preprocessor as CPPFLAGS solved similar issue in my M1
-DPNG_ARM_NEON_OPT=0
Pass the value as env var with key CPPFLAGS to relevant service.
Provided the base image includes the target architecture, another option that might work in your case is using Docker's built-in TARGETARCH build arg. This works for me on macOS M1.
FROM ubuntu:focal
ARG TARGETARCH
RUN apt update; apt install -y curl jq build-essential python3.8 python3-pip docker-compose jsonnet bison mercurial
RUN ln -s /usr/bin/python3.8 /usr/bin/python
RUN curl -OL https://golang.org/dl/go1.17.linux-${TARGETARCH}.tar.gz; mkdir /etc/golang; tar -xvzf go1.17.linux-${TARGETARCH}.tar.gz -C /etc/golang; ln -s /etc/golang/go/bin/go /usr/bin/go; rm -f go1.17.linux-${TARGETARCH}.tar.gz
RUN GO111MODULE="on" go get github.com/jsonnet-bundler/jsonnet-bundler/cmd/jb; ln -s $(go env GOPATH)/bin/jb /usr/bin/jb
WORKDIR /workspace

Why can't I get ocaml/opam:ubuntu-16.04_ocaml-4.03.0 docker Image?

I was trying to run:
docker build -f hol-light/Dockerfile_check_proofs --ulimit stack=1000000000 --tag check_proofs hol-light/
but I get the error:
Sending build context to Docker daemon 48.9MB
Step 1/16 : FROM ocaml/opam:ubuntu-16.04_ocaml-4.03.0
pull access denied for ocaml/opam, repository does not exist or may require 'docker login': denied: requested access to the resource is denied
Why?
The docker file is: https://github.com/brain-research/hol-light/blob/master/Dockerfile_check_proofs
FROM ocaml/opam:ubuntu-16.04_ocaml-4.03.0
WORKDIR /home/opam/
SHELL ["/bin/bash", "-c"]
ENV PATH="/home/opam/.opam/4.03.0/bin:${PATH}"
### Install num
RUN opam install num
### Install campl5
RUN git clone --depth 1 -b rel617 https://github.com/camlp5/camlp5
RUN cd camlp5 &&\
./configure &&\
make world.opt &&\
make install &&\
# meta/Makefile in camlp5 skips these files which we need, so copy them
# manually.
cp {main/pcaml,main/quotation,etc/pa_reloc,meta/q_MLast}.{cmi,cmx,o} `camlp5 -where`
### Install grpc
RUN sudo apt-get update &&\
sudo apt-get install -y build-essential autoconf libtool pkg-config clang libc++-dev
RUN git clone -b 'v1.17.1' --recurse-submodule --depth 1 https://github.com/grpc/grpc
RUN sudo make -C grpc install-headers_c install-static_c install-pkg-config_c\
install-headers_cxx install-static_cxx install-pkg-config_cxx\
install-plugins
RUN sudo make -C grpc/third_party/protobuf install
### Install farmhash
RUN git clone --depth 1 https://github.com/google/farmhash &&\
cd farmhash &&\
./configure CXXFLAGS="-DNAMESPACE_FOR_HASH_FUNCTIONS=farmhash"
RUN sudo make -C farmhash install
### Build binaries
COPY --chown=opam:0 . src/
RUN make -C src check_proofs
CMD ["./src/check_proofs"]
crossposted:
https://forums.docker.com/t/why-cant-i-get-ocaml-opam-ubuntu-16-04-ocaml-4-03-0-docker-image/84351
https://hub.docker.com/r/ocaml/opam/ hasn't been updated for 2 years and says:
At some point in the future, the tags in this repository will be deleted
This deletion is currently in progress (in fact, it has been deleting for more than a week now).
The ocaml/opam (opam 1) images generally aren't useful now because they don't work with the current opam-repository.
There are two alternatives you can use:
ocaml/opam2 contains opam 2 images. e.g. ocaml/opam2:ubuntu-16.04-ocaml-4.03
ocurrent/opam is also opam 2, but contains much smaller images (with only one version of the compiler per image). e.g. ocurrent/opam:ubuntu-16.04-ocaml-4.03
However, this repository is only temporary. It will replace ocaml/opam once Hub finishes deleting that...

Docker: Reverse Engineering of an Image

When we use Docker it's very easy push and pull image in a public repository in our https://hub.docker.com but this repository it's free only for public image(only one can be private).
Currently it's possible to execute a reverse engineering of a public image in repository and read the source code of project ?
You can check how an image was created using docker history <image-name> --no-trunc
Update:
Check dive which is a very nice tool that allows you to views image layers.
As yamenk said docker history is the key to this.
As https://github.com/CenturyLinkLabs/dockerfile-from-image is broken, you can use recent
https://hub.docker.com/r/dduvnjak/dockerfile-from-image/
Extract from the site
Note that the script only works against images that exist in your local image
repository (the stuff you see when you type docker images). If you want to
generate a Dockerfile for an image that doesn't exist in your local repo
you'll first need to docker pull it.
For example, you can run it agains itself, to see the code
$ docker run --rm -v /run/docker.sock:/run/docker.sock centurylink/dockerfile-from-image ruby
FROM buildpack-deps:latest
RUN useradd -g users user
RUN apt-get update && apt-get install -y bison procps
RUN apt-get update && apt-get install -y ruby
ADD dir:03090a5fdc5feb8b4f1d6a69214c37b5f6d653f5185cddb6bf7fd71e6ded561c in /usr/src/ruby
WORKDIR /usr/src/ruby
RUN chown -R user:users .
USER user
RUN autoconf && ./configure --disable-install-doc
RUN make -j"$(nproc)"
RUN make check
USER root
RUN apt-get purge -y ruby
RUN make install
RUN echo 'gem: --no-rdoc --no-ri' >> /.gemrc
RUN gem install bundler
ONBUILD ADD . /usr/src/app
ONBUILD WORKDIR /usr/src/app
ONBUILD RUN [ ! -e Gemfile ] || bundle install --system
You can use laniksj/dfimage to reverse engineering of an image.
For example:
# docker run -v /var/run/docker.sock:/var/run/docker.sock laniksj/dfimage <YOUR_IMAGE_ID>
FROM node:12.4.0-alpine
RUN /bin/sh -c apk update
RUN /bin/sh -c apk -Uuv add groff less python py-pip
RUN /bin/sh -c pip install awscli
RUN /bin/sh -c apk --purge -v del py-pip
RUN /bin/sh -c rm /var/cache/apk/*
RUN /bin/sh -c apk add --no-cache curl
ADD dir:4afc740ff29e4a32a34617d2715e5e5dc8740f357254bc6d3f9362bb04af0253 in /app
COPY file:b57abdb61ae72f3a25be67f719b95275da348f9dfb63fb4ff67410a595ae1dfd in /usr/local/bin/
WORKDIR /app
RUN /bin/sh -c npm install
ENTRYPOINT ["docker-entrypoint.sh"]
CMD ["node" "app.js"]
dfimage and dockerfile-from-image are broken
dedockify works
imageName=ruby:latest
docker pull $imageName
docker images # -> get imageId
imageId=xxxxxxxxxxxx
# write to Dockerfile
docker run --rm -v /var/run/docker.sock:/var/run/docker.sock mrhavens/dedockify $imageId >Dockerfile

Perl installation issue using Docker

I am trying to build the docker image with perl installation.
Dockerfile:
FROM amazonlinux
WORKDIR /shared
RUN yum -y install gcc
ADD http://www.cpan.org/src/5.0/perl-5.22.1.tar.gz /shared
RUN tar -xzf perl-5.22.1.tar.gz
WORKDIR /shared/perl-5.22.1
RUN ./Configure -des -Dprefix=/opt/perl-5.22.1/localperl
RUN make
RUN make test
RUN make install
all these steps are executed i am can see it executing the make, make test and make install commands but when i do :
docker run -it testsh /bin/bash
Error:
when I check perl -v it says command not found.
and I need to go the perl directory
'cd perl-5.22.1' and run 'make install' again then perl -v works
But I want the perl installation to work when I build it with docker image. can anyone tell me what is going wrong here?
perl was indeed installed, just wasn't added to the path.
export PATH=$PATH:/shared/perl-5.22.1 should do it -- but of course, you'd want to add a PATH update in the Dockerfile.
At first glance I thought that when you run make install second time, it adds perl's bin directory to PATH env, but when I compared output of env before and after make install it showed the same PATH variable content.
The reason you getting perl -v working after make install in running container is that make install puts perl binary to /usr/bin/perl. I don't know why it works such way, but it is just as it is. Also, it's almost useless to store sources inside of your image.
Anyway, I agree with #belwood suggestion about adding your perl's bin directiry to PATH environment variable. I just wanna correct the path: /opt/perl-5.22.1/localperl/bin
You need to add it in your Dockerfile (basically I've rewritten your file to make it produce more efficient image), for example:
FROM amazonlinux
RUN mkdir -p /shared/perl-5.22.1
WORKDIR /shared/perl-5.22.1
RUN yum -y install gcc \
&& curl -SL http://www.cpan.org/src/5.0/perl-5.22.1.tar.gz -o perl-5.22.1.tar.gz \
&& tar --strip-components=1 -xzf perl-5.22.1.tar.gz \
&& rm perl-5.22.1.tar.gz \
&& ./Configure -des -Dprefix=/opt/perl-5.22.1/localperl \
&& make -j $(nproc) \
&& make -j $(nproc) test \
&& make install \
&& rm -fr /shared/perl-5.22.1 /tmp/*
ENV PATH="/opt/perl-5.22.1/localperl/bin:$PATH"
WORKDIR /root
CMD ["perl","-de0"]
When you simply run container with this image, you'll immediately get into perl's shell. If you need bash, then use docker run -it --rm amazon-perl /bin/bash
It would be also good to look at Environment replacement section in the Dockerfile reference documentation, just to figure out how things work. For example, it isn't a best pratice to have that many RUN lines in your Dockerfile because of the RUN instruction will execute commands in a new layer on top of the current image and commit the results. So you'll get many unnecessary layers.

Some RUNs won't work on docker but will when inside a container

I've got a Dockerfile for some lua and torch related tasks and I'm trying to install some rocks using luarocks.
FROM ubuntu:14.04
RUN rm /bin/sh && ln -s /bin/bash /bin/sh
RUN apt-get update -y
RUN apt-get install -y curl git
RUN curl -s https://raw.githubusercontent.com/torch/ezinstall/master/install-deps | bash
RUN git clone https://github.com/torch/distro.git ~/torch --recursive
RUN cd ~/torch; ./install.sh
RUN source ~/.bashrc
RUN luarocks install nngraph
RUN luarocks install optim
RUN luarocks install nn
RUN luarocks install cltorch
RUN luarocks install clnn
docker build runs fine up until the first luarocks call: RUN luarocks install nngraph at which point it stops and throws the error:
/bin/sh: luarocks: command not found
If I comment out the luarocks lines, the build runs fine. Using that image, I can create a container and using bash, run luarocks as expected.
Of course, I don't particularly want to have to do this every time I start up the container, so I'm wondering if there's anything I can do to make this work. I have a feeling that this problem has something to do with the line RUN rm /bin/sh && ln -s /bin/bash /bin/sh but I need that to be able to run the line RUN source ~/.bashrc.
Thanks.
Each RUN command runs on its own shell and a new layer is committed.
From the Docker documentations:
RUN (the command is run in a shell - /bin/sh -c - shell
form)
So when you run luarocks install <app> it will not be the same shell you source your profile.
You have to provide the full path for luarocks to run. See below a modified Dockerfile I successfully run:
FROM ubuntu:14.04
RUN rm /bin/sh && ln -s /bin/bash /bin/sh
RUN apt-get update -y
RUN apt-get install -y curl git
RUN curl -s https://raw.githubusercontent.com/torch/ezinstall/master/install-deps | bash
RUN git clone https://github.com/torch/distro.git ~/torch --recursive
RUN cd ~/torch; ./install.sh
RUN source ~/.bashrc
RUN /root/torch/install/bin/luarocks install nngraph
RUN /root/torch/install/bin/luarocks install optim
RUN /root/torch/install/bin/luarocks install nn
RUN /root/torch/install/bin/luarocks install cltorch
RUN /root/torch/install/bin/luarocks install clnn
For more details see docker RUN documentation here.
As Alex da Silva pointed it out, sourcing .bashrc happens in another shell in your Dockerfile.
You could also try that to have your luarocks commands executed in the same shell with your sourced bashrc:
...
RUN source ~/.bashrc && luarocks install nngraph
RUN source ~/.bashrc && luarocks install optim
RUN source ~/.bashrc && luarocks install nn
RUN source ~/.bashrc && luarocks install cltorch
RUN source ~/.bashrc && luarocks install clnn

Resources