I have a machine which has no internet connectivity and no access to any docker repository (so no image pulls are possible). I want to install memcached and I have the .rpm file available.
When I want to install on the host machine I execute the command rpm -ivh memcached-1.4.15-10.el7_3.1.x86_64.rpm . But I suppose that is because the rpm package manager comes pre-installed on the host OS.
In the docker container I load the .rpm file and in the dockerfile I include the command RUN rpm -ivh memcached-1.4.15-10.el7_3.1.x86_64.rpm. After that I get the following error:
/bin/sh: 1: rpm: not found
The command '/bin/sh -c rpm -ivh /home/memcached-1.4.15-10.el7_3.1.x86_64.rpm' returned a non-zero code: 127
I suppose that is because in a docker container the OS has the bare minimum installation. So how do I install the rpm package manager inside the container without internet connection? Is there an installable file for it.
I understand it is not the best practice to not use a central repository for images. Want to know if installing without internet is even possible?
I am creating the container on a CentOS machine right now. And following is the dockerfile:
FROM microsoft/dotnet:2.0-runtime
WORKDIR /home
#mempkgtest folder contains the .rpm file
COPY ${source:-mempkgtest} .
RUN rpm -ivh /home/memcached-1.4.15-10.el7_3.1.x86_64.rpm
ENTRYPOINT ["dotnet", "--info"]
Docker image microsoft/dotnet is build on top of buildpack-deps:jessie-scm (from here) which is build on top debian:jessie (from here and here).
debian does not use rpm package manager, it uses deb format for packages. /bin/sh kindly informs you, that is hasn't found rpm manager by saying /bin/sh: 1: rpm: not found. You can read how to install rpm packages on debian here.
Anyway, why don't you use deb file and dpkg package manager? You can find memcached pacakge for debian jessie here. You can do smth like this in your dockerfile:
ADD http://ftp.us.debian.org/debian/pool/main/m/memcached/memcached_1.4.21-1.1+deb8u1_amd64.deb
RUN dpkg -i memcached_1.4.21-1.1+deb8u1_amd64.deb
And remember, you will need to copy dependencies too.
Why don't you make your docker image on machine with internet access, than export docker image using docker export and then copy and import it on your destination machine? Such way is simpler and apt-get will resolve and install all memcached dependencies for you.
Related
I need to install python 3 on my virtual machine (I have python 2.7) but I don't have access to internet from my VM. Is there any way to do that without using internet I have access to a private gitlab repository and private dokcer hub.
Using GitLab
Ultimately, you can put whatever resources you need to install Python3 directly in GitLab.
For example, you could use the generic packages registry to upload the files you need and download them from GitLab in your VM. For example, you can redistribute the files from python.org/downloads this way.
If you're using a debian-based Linux distribution like Ubuntu, you could even provide the necessary packages in the GitLab debian registry (disabled by default, but can be enabled by an admin) and just use your package manager like apt install python3-dev after configuring your apt lists to point to the gitlab debian repo.
Using docker
If you have access to dockerhub, technically you can access files from docker images as well. Here I'll assume you're using ubuntu or some debian-based distribution, but the same principle applies for any OS.
Suppose you build an image:
FROM ubuntu:<a tag that matches your VM version>
# downloads all the `.deb` files you need to install python3
RUN apt update && apt install --download-only python3-dev
You can push this image to your docker registry
Then on your VM, you can pull this image and extract the necessary install files from /var/cache/apt/archives/*.deb in the image then install using dpkg
Extract files from the image (in this case, to a temp directory)
image=myprivateregistry.example.com/myrepo/myimage
source_path=/var/cache/apt/archives
destination_path=$(mktemp -d)
docker pull "$image"
container_id=$(docker create "$image")
docker cp "$container_id:$source_path" "$destination_path"
docker rm "$container_id"
Install python3 using dpkg:
dpkg --force-all -i "${destination_path}/*.deb"
I am trying to install dotnet-sdk-3.0 on linux AMI 2 ec2 instance (c6g). I am new to linux so tried couple of commands but nothing seems working for me. I tried below.
sudo rpm -Uvh https://packages.microsoft.com/config/centos/7/packages-microsoft-prod.rpm
sudo yum install dotnet-sdk-3.1
sudo yum install dotnet-sdk-3.0
When tried above i am getting below error.
[ec2-user#ip-0-0-0-0 console]$ sudo yum install dotnet-sdk-3.0
Loaded plugins: extras_suggestions, langpacks, priorities, update-motd
amzn2-core | 3.7 kB 00:00
amzn2extra-docker | 3.0 kB 00:00
amzn2extra-nginx1.12 | 1.3 kB 00:00
packages-microsoft-com-prod | 3.0 kB 00:00
packages-microsoft-com-prod/primary_db | 288 kB 00:00
No package dotnet-sdk-3.0 available.
Error: Nothing to do
Then i tried
mkdir -p "$HOME/dotnet" && tar zxf dotnet-sdk-3.0.100-linux-x64.tar.gz -C "$HOME/dotnet"
export DOTNET_ROOT=$HOME/dotnet
export PATH=$PATH:$HOME/dotnet
After this tried the dotnet command but got the error. dotnet: command not found
Finally tried below:
mkdir -p $HOME/dotnet && tar zxf dotnet-sdk-3.1.302-linux-arm64.tar.gz -C $HOME/dotnet
export DOTNET_ROOT=$HOME/dotnet
export PATH=$PATH:$HOME/dotnet
dotnet
dotnet --list-sdk
when i run this, i got below error
[ec2-user#ip-0-0-0-0 home]$ dotnet --list-sdk
Process terminated. Couldn't find a valid ICU package installed on the system. Set the configuration flag System.Globalization.Invariant to true if you want to run with no globalization support.
at System.Environment.FailFast(System.String)
at System.Globalization.GlobalizationMode.GetGlobalizationInvariantMode()
at System.Globalization.GlobalizationMode..cctor()
at System.Globalization.CultureData.CreateCultureWithInvariantData()
at System.Globalization.CultureData.get_Invariant()
at System.Globalization.CultureInfo..cctor()
at System.String.ToLowerInvariant()
at Microsoft.DotNet.PlatformAbstractions.RuntimeEnvironment.GetArch()
at Microsoft.DotNet.PlatformAbstractions.RuntimeEnvironment..cctor()
at Microsoft.DotNet.PlatformAbstractions.RuntimeEnvironment.GetRuntimeIdentifier()
at Microsoft.DotNet.Cli.MulticoreJitProfilePathCalculator.CalculateProfileRootPath()
at Microsoft.DotNet.Cli.MulticoreJitActivator.StartCliProfileOptimization()
at Microsoft.DotNet.Cli.MulticoreJitActivator.TryActivateMulticoreJit()
at Microsoft.DotNet.Cli.Program.Main(System.String[])
Aborted
also tried to run the dotnet <dotnet-project.dll> and received this error.
Failed to load â–’râ–’), error: /home/ec2-user/dotnet/shared/Microsoft.NETCore.App/3.0.0/libhostpolicy.so: cannot open shared object file: No such file or directory
An error occurred while loading required library libhostpolicy.so from [/home/ec2-user/dotnet/shared/Microsoft.NETCore.App/3.0.0]
[ec2-user#ip-0-0-0-0 console]$
I have followed the microsoft document as well.
https://learn.microsoft.com/en-us/dotnet/core/install/linux-centos
Nothing seems working for me. Can someone please help me here, i am stuck from last 2 days.
Thanks in advance.
PS: I am completely newbie to linux.
I was able to get this to work on Amazon Linux 2 ARM64 with the following steps:
Download and build a recent version of International Components for Unicode. Amazon Linux 2 has version 50 available in the package manager, but dotnet looks for version 55 or later. When Amazon Linux updates this, this step won't be necessary anymore and can be replaced with sudo yum install icu.
Install gcc and python3 because they'll be needed for building libicu. It takes a few minutes to build the library. Amazon Linux 2 recently added libicu60 to the package manager, so you can simply install it with yum:
sudo yum -y install libicu60
Follow the instructions from Microsoft to install the dotnet sdk. Download the dotnet sdk from the Microsoft Website. You may need to download from this link to get the latest version, but the link below worked for me.
cd ~
wget https://download.visualstudio.microsoft.com/download/pr/5ee48114-19bf-4a28-89b6-37cab15ec3f2/f5d1f54ca93ceb8be7d8e37029c8e0f2/dotnet-sdk-3.1.302-linux-arm64.tar.gz
mkdir -p $HOME/dotnet && tar zxf ~/dotnet-sdk-3.1.302-linux-arm64.tar.gz -C $HOME/dotnet
export DOTNET_ROOT=$HOME/dotnet
export PATH=$PATH:$HOME/dotnet
Test that it's working
[ec2-user#ip-172-31-69-243 ~]$ dotnet --list-sdks
3.1.302 [/home/ec2-user/dotnet/sdk]
You are running this on arm64/aarch64. It's a relatively new architecture. It's also incompatible with the Intel 64-bit architecture (x86_64 or x64). So you need to watch out for that.
Installing via RPM
Edit: So, this is just not going to work if you want to use RPM packages.
Quoting https://learn.microsoft.com/en-us/dotnet/core/install/linux-centos:
Package manager installs are only supported on the x64 architecture. Other architectures, such as ARM, must manually install the .NET Core SDK or .NET Core Runtime. For more information, see the manually install section below.
You are using aarch64/arm64. You are not using x64, so this is not going to work.
You need to use the tarball installation method.
Out of date suggestions:
I am trying to install dotnet-sdk-3.0 on linux AMI 2 ec2 instance (c6g).
sudo rpm -Uvh https://packages.microsoft.com/config/centos/7/packages-microsoft-prod.rpm
You are running Amazon Linux 2, right? As the URL here says, this is for CentOS 7. It may (or it may not) work on your Linux distribution. Anyway, try it out.
$ sudo yum install dotnet-sdk-3.0
No package dotnet-sdk-3.0 available.
Error: Nothing to do
The error says that it can't find this package. Maybe a package with this name doesn't exist? Maybe you are using the wrong name? Try using yum list to find the correct name:
sudo yum list 'dotnet-sdk*'
It should show you a list of packages, including names like dotnet-sdk-3.0.103. You can install that package by name, then:
sudo yum install dotnet-sdk-3.0.103
If that doesn't work, try another package name from yum list and try installing that.
Installing manually
Then i tried
mkdir -p "$HOME/dotnet" && tar zxf dotnet-sdk-3.0.100-linux-x64.tar.gz -C "$HOME/dotnet"
export DOTNET_ROOT=$HOME/dotnet
export PATH=$PATH:$HOME/dotnet
After this tried the dotnet command but got the error. dotnet: command not found
You are running an aarch64 machine. You need to use the arm64 tarball, not the x64 tarball. The x64 tarball is for an Intel processor. It will not work on an ARM processor.
That's surprising. Let me break down what this set of steps is doing:
mkdir -p "$HOME/dotnet" creates a directory named dotnet in your home directory
tar xf ... extracts the dotnet SDK tarball in the dotnet directory you created in step 1
export DOTNET_ROOT=$HOME/dotnet defines an environment variable DOTNET_ROOT. .NET Runtime needs it; I am a bit fuzzy myself on why
export PATH=$PATH:$HOME/dotnet adds the directory you installed the .NET SDK into to the environment variable PATH. PATH is a list of locations that the OS uses to search for a command that you enter. For example, when you type dotnet in the command line it searches for dotnet executable (think dotnet.exe on Windows) in this list of directories.
So let's try and debug it one by one:
Does the directory dotnet exist in your main home directory (aka $HOME)? Can you cd ~/dotnet? Does that work?
After you extract the tarball, do you see a file named dotnet in the dotnet directory in your $HOME? Does ls $HOME/dotnet/dotnet work? What does it show you?
What does echo $PATH show you? Does it include that dotnet directory in the value?
If you run which dotnet, does it find the dotnet executable in your main $HOME directory?
Running the SDK
when i run this, i got below error
[ec2-user#ip-0-0-0-0 home]$ dotnet --list-sdk
Process terminated. Couldn't find a valid ICU package installed on the system. Set the configuration flag System.Globalization.Invariant to true if you want to run with no globalization support
The error includes this phrase: Couldn't find a valid ICU package installed on the system.
It really means that. You need to install the ICU package for your Linux distribution:
sudo yum install libicu
And then try running dotnet --list-sdk again.
Error Running dll
Failed to load â–’râ–’), error: /home/ec2-user/dotnet/shared/Microsoft.NETCore.App/3.0.0/libhostpolicy.so: cannot open shared object file: No such file or directory
An error occurred while loading required library libhostpolicy.so from [/home/ec2-user/dotnet/shared/Microsoft.NETCore.App/3.0.0]
This is strange. It says it can't find a file that should be part of the .NET Core installation.
What does dotnet --list-runtimes say? Does it show the 3.0.0 runtime installed? If not, that means your installation is messed up. You should probably install .NET Core 3.0 again. (Or better yet, install 3.1 because 3.0 has been end-of-life'd).
Does the file /home/ec2-user/dotnet/shared/Microsoft.NETCore.App/3.0.0/libhostpolicy.so exist? If it doesn't it's the same problem as above: your installation is messed up.
What does file /home/ec2-user/dotnet/shared/Microsoft.NETCore.App/3.0.0/libhostpolicy.so say? Is it an ELF 64-bit LSB shared object?
The output is: ELF 64-bit LSB shared object, x86-64
This is a x86-64 file! In other words, you have (somehow) installed an linux-x64 (Intel 64-bit architecture) runtime. Not too surprisingly, it doens't work on the ARM 64 bit architecture. You need to delete this and re-install the SDK. I suggest just blowing away your current installation (rm -rf $HOME/dotnet) and install the linux-arm64 SDK again.
History:
My docker build file worked for years, without any problem, on my Linux Mint VM. When I needed to recreate the VM, I installed everything again, including docker.io.
I'm taking a beating with this error. I already verified that the final file is inside the docker image, but when I try to copy it to a directory external to the container, it says that it does not exist.
I followed the guidelines at Exploring Docker container's file system and verified that the file was in fact in the container.
Environment:
Linux Mint 19 (Tricia)
Docker installed by snap
Command:
docker cp {CONTAINER_ID}:/container_path /local_path
Problem:
stat /container_path: no such file or directory
The solution was simply to uninstall the docker by snap and install it again by apt. This solution still lacks more information, as it is not known if the problem was really caused by the version of the docker installed by snap.
sudo snap remove docker
sudo apt install docker.io
I am running Jenkins in docker from official docker hub .
I created job which runs my own shell script, however I see some binaries
are missing in docker e.g.file command.
They mention on docker hub that one can install additional binaries over Ubuntu's aptitude however I don't know which package to install to get e.g file command working.
Unless Ubuntu did something different than the base Debian environment, file is included in the file package.
apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -f file
i am trying to install a package using rpm, for which i have created a different database using rpmdb --initdb --dbpath $HOME/myrpmdb and specifying that path in the --dbpath while insatlling.
The error i am getting is
root#jason:su rpm --dbpath $HOME/myrpmdb -ivh XXX.rpm
error: Failed dependencies:
/bin/sh is needed by XXX
and /bin/sh exist
Is there some way that i can specify this to the installer?
Is there something i missed during the creation of different rpm database cause of which now it not able to fine the default commands?
Right way install rpm via alien:
alien -i my_rpm_name.rpm
If alien is not installed:
sudo apt-get install alien
error: Failed dependencies:
/bin/sh is needed by XXX
This happens because bash rpm is not available in the new rpmdb path. You would have to use following to get the rpm installed.
# rpm --dbpath $HOME/myrpmdb --nodeps -ivh XXX.rpm
Because you've told it to use a completely different database, rpm doesn't look at the system database which is where /bin/sh would be found.
Why use a different rpm database? There should be a really good reason to go this route...
If you must have a private rpm db, you can copy the system db into your private one
cp /var/lib/rpm/* ~/myrpmdb/.
installing with --nodeps may be the easiest way... copying the rpm db is really sub-optimal, it will soon be out of date and not reflect the state of the system.