I'm installing rvm as part of script. After installed I need to run sudo gem but to do this I first need to source the script.
The recommended way is to use source /home/$USER/.rvm/scripts/rvm, but this is not possible with sudo as described here.
I read that source was equivilent to ./$FILE_NAME but ./home/$USER/.rvm/scripts/rvm didn't solve my issue.
How can I source rvm so ruby and gem can be accessed via sudo?
Restarting or logging out is not possible as this is all done in a single shell script.
I initially tried referencing rvm and gem by the full path to the file. This got me a step further but led to other issues. From what I've read it seems sudo use with rvm isn't recommended.
I decided to not use rvm and download the latest ruby from their official downloads page instead.
My install_ruby.sh script: (run from folder to install in)
# Download ruby
wget https://cache.ruby-lang.org/pub/ruby/2.4/ruby-2.4.1.tar.gz -P /tmp
# Get SHA256 sum
SHA="$(sha256sum /tmp/ruby-2.4.1.tar.gz | cut -d ' ' -f1)"
# Exit if SHA256 doesn't match expected
[ "$SHA" != "a330e10d5cb5e53b3a0078326c5731888bb55e32c4abfeb27d9e7f8e5d000250" ] && exit;
# Extract and install tar
tar -zxvf /tmp/ruby-2.4.1.tar.gz -C /opt
/opt/ruby-2.4.1/configure
make
sudo make install
Related
I am running a debian VM Instance using GCP Compute Engine and I have added a automation script to be executed on startup.
There are few tools which will be downloaded on startup. Only issue is, everything is getting downloaded in / directory.
I need to download everything in $HOME directory.
Different ways I have tried
#!/bin/bash
set -x
cd $HOME
mkdir $HOME/test
cd $HOME/test
apt install wget -y
wget https://download.java.net/openjdk/jdk11/ri/openjdk-11+28_linux-x64_bin.tar.gz
#!/bin/bash
set -x
source $HOME
mkdir $HOME/something
#!/bin/bash
set -x
cd $HOME
mkdir $HOME/something
exec bash
Still it is downloaded in / directory. What else can be done here?
You are trying to make 2 things : install wget package and download another one.
Why don't you tried to install wget manually ?
apt-get install wget
You have then to store the full path for your script, and download the package needed it it. Try this :
#!/bin/bash
homePath=$HOME
mkdir $HOME/test
wget https://download.java.net/openjdk/jdk11/ri/openjdk-11+28_linux-x64_bin.tar.gz -P $homePath/test/
How do I install the anaconda / miniconda without prompts on Linux command line?
Is there a way to pass -y kind of option to agree to the T&Cs, suggested installation location etc. by default?
can be achieved by bash miniconda.sh -b (thanks #darthbith)
The command line usage for this can only be seen with -h flag but not --help, so I missed it.
To install the anaconda to another place, use the -p option:
bash anaconda.sh -b -p /some/path
AFAIK pyenv let you install anaconda/miniconda
(after successful instalation)
pyenv install --list
pyenv install miniconda3-4.3.30
For a quick installation of miniconda silently I use a wrapper
script script that can be executed from the terminal without
even downloading the script. It takes the installation destination path
as an argument (in this case ~/miniconda) and does some validation too.
curl -s https://gist.githubusercontent.com/mherkazandjian/cce01cf3e15c0b41c1c4321245a99096/raw/03c86dae9a212446cf5b095643854f029b39c921/miniconda_installer.sh | bash -s -- ~/miniconda
Silent installation can be done like this, but it doesn't update the PATH variable so you can't run it after the installation with a short command like conda:
cd /tmp/
curl -LO https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh -b -u
Here -b means batch/silent mode, and -u means update the existing installation of Miniconda at that path, rather than failing.
You need to run additional commands to initialize PATH and other shell init scripts, e.g. for Bash:
source ~/miniconda3/bin/activate
conda init bash
I installed leiningen on fedora, I followed the instruction:
download the script
Paste the script(via command line) to ~/bin (PATH)
Execute the script
Run lein(I had to did it with sudo) to self-install
but every time I want to run the "lein" command, I have to do it with "sudo".
How can i fix this? or what can i do to fix this?
Note: I installed leiningen at /bin but when i cd ~/bin as the installation guide said i get and error about the folder(not exist).
You installed it in /bin and ran it with sudo to install the lein jars initially, which means they are owned (and probably only readable) by root. You should install the script at ~/bin instead. You can fix it like this:
sudo rm /bin/lein
sudo rm ~/.lein
mkdir ~/bin
cd ~/bin
wget https://raw.githubusercontent.com/technomancy/leiningen/stable/bin/lein
export PATH="${HOME}/bin:${PATH}"
lein
You should also add ${HOME}/bin to your $PATH. If you are using bash, add this to ~/.bashrc:
export PATH="${HOME}/bin:${PATH}"
You can do that using echo:
echo 'export PATH="${HOME}/bin:${PATH}"' >> ~/.bashrc
anybody is familiar with the etcd project? Or we'd better forget the project when talk about this issue. The issue is
$ build
ln: `gopath/src/github.com/coreos/etcd': cannot overwrite directory
when exec the build shell
and the content is:
#!/bin/sh -e
if [ ! -h gopath/src/github.com/coreos/etcd ]; then
mkdir -p gopath/src/github.com/coreos/
ln -s ../../../.. gopath/src/github.com/coreos/etcd
fi
export GOBIN=${PWD}/bin
export GOPATH=${PWD}/gopath
export GOFMTPATH="./bench ./config ./discovery ./etcd ./error ./http ./log main.go ./metrics ./mod ./server ./store ./tests"
# Don't surprise user by formatting their codes by stealth
if [ "--fmt" = "$1" ]; then
gofmt -s -w -l $GOFMTPATH
fi
go install github.com/coreos/etcd
go install github.com/coreos/etcd/bench
Some addition:
My system is windows 7
I run the shell on git bash.
issue reproduce:
step1: open the git bash
step2: git clone git#github.com:coreos/etcd.git
step3: cd etcd
step4: build
As mentioned in "Git Bash Shell fails to create symbolic links" (since you are using the script in a git bash on Windows 7)
the ln that shipped with msysGit simply tries to copy its arguments, rather than fiddle with links. This is because links only work (sort of) on NTFS filesystems, and the MSYS team didn't want to reimplement ln.
A workaround is to run mklink from Bash.
This also allows you to create either a Symlink or a Junction.
So 'ln' wouldn't work as expected by default, in the old shell that ships with Git for Windows.
Here's solution. Tbh it is a workaround, but since you're on Windows, I don't see another way.
Start a command line, and enter there to the directory with the script. There should be a path gopath/src/github.com/coreos/ (if no such a path, you must create it). Next issue a command
mklink /D "gopath/src/github.com/coreos/etcd" "../../../../"
Next you should edit the build script to delete a lines with creation symlink and a directory. E.g.
#!/bin/sh -e
export GOBIN=${PWD}/bin
export GOPATH=${PWD}/gopath
export GOFMTPATH="./bench ./config ./discovery ./etcd ./error ./http ./log main.go ./metrics ./mod ./server ./store ./tests"
# Don't surprise user by formatting their codes by stealth
if [ "--fmt" = "$1" ]; then
gofmt -s -w -l $GOFMTPATH
fi
go install github.com/coreos/etcd
go install github.com/coreos/etcd/bench
Note, that I am just removed 4 lines of code. Next you run the script, and this should work.
You shouldn't be using git clone and the build sh script. Use the go get command. For example, on Windows 7,
Microsoft Windows [Version 6.1.7601]
C:\>set gopath
GOPATH=C:\gopath
C:\>go version
go version go1.3 windows/amd64
C:\>go get -v -u github.com/coreos/etcd
github.com/coreos/etcd (download)
github.com/coreos/etcd/third_party/bitbucket.org/kardianos/osext
github.com/coreos/etcd/pkg/strings
github.com/coreos/etcd/error
github.com/coreos/etcd/third_party/github.com/coreos/go-etcd/etcd
github.com/coreos/etcd/http
github.com/coreos/etcd/third_party/github.com/coreos/go-log/log
github.com/coreos/etcd/third_party/github.com/rcrowley/go-metrics
github.com/coreos/etcd/mod/dashboard/resources
github.com/coreos/etcd/log
github.com/coreos/etcd/third_party/github.com/gorilla/context
github.com/coreos/etcd/third_party/github.com/gorilla/mux
github.com/coreos/etcd/mod/dashboard
github.com/coreos/etcd/discovery
github.com/coreos/etcd/pkg/btrfs
github.com/coreos/etcd/pkg/http
github.com/coreos/etcd/third_party/code.google.com/p/gogoprotobuf/proto
github.com/coreos/etcd/mod/leader/v2
github.com/coreos/etcd/mod/lock/v2
github.com/coreos/etcd/metrics
github.com/coreos/etcd/third_party/github.com/mreiferson/go-httpclient
github.com/coreos/etcd/mod
github.com/coreos/etcd/third_party/github.com/BurntSushi/toml
github.com/coreos/etcd/third_party/github.com/goraft/raft/protobuf
github.com/coreos/etcd/third_party/github.com/goraft/raft
github.com/coreos/etcd/store
github.com/coreos/etcd/server/v1
github.com/coreos/etcd/server/v2
github.com/coreos/etcd/store/v2
github.com/coreos/etcd/server
github.com/coreos/etcd/config
github.com/coreos/etcd/etcd
github.com/coreos/etcd
C:\>
I am new to linux and am trying to install nodejs latest version with binaries. The solutions I have looked up suggest the installation using apt-get on some private repositories(PPA), which I do not want to do.
So I ran the following commands:
wget http://nodejs.org/dist/v0.10.26/node-v0.10.26-linux-x64.tar.gz
tar -zxvf node-v0.10.26-linux-x64.tar.gz
mv node-v0.10.26-linux-x64 node-v0.10.26
sudo cp -r node-v0.10.26 /usr/local/src
After this, I don't really know what to do. I read an article which suggested created symbolic links, which I am kind of scared to mess up with without knowing the details.
Could you please give me a set of commands to run after this in order to install node with npm? I guess npm should be a part of this binary version.
The best way to install Node.js and have the latest version (or any other version that you prefer, be it LTS or "current") is to download the official binary bundle and uncompress it. A neat way to do it:
# Use version 0.10.26
$ NODE_VERSION="v0.10.26"
# To use a newer version, for example 6.10.3, use instead:
$ NODE_VERSION="v6.10.3"
$ curl -LO http://nodejs.org/dist/$NODE_VERSION/node-$NODE_VERSION-linux-x64.tar.gz
$ tar xzf node-$NODE_VERSION-linux-x64.tar.gz
$ sudo cp -rp node-$NODE_VERSION-linux-x64 /usr/local/
$ sudo ln -s /usr/local/node-$NODE_VERSION-linux-x64 /usr/local/node
The basic idea is to move all the contents of the archive into /usr/local, then create a symlink in /usr/local/node pointing to the most recent version.
For enabling the use of the "node" executable from the command line without referencing the full path (/usr/local/node/bin/node), add /usr/local/node/bin to your $PATH (usually this involves altering the ~/.bash_profile or ~/.profile file; there's plenty of docs for how to do this).
If you need to update Node.js (suppose it's version 7.10.0), then, just extract the tarball in /usr/local and update the symbolic link so it points to the new one. You can then optionally remove the old folder.
May 2017 update
As of the "Creators Update", the commands above can now work also on Windows 10 using the "Windows Subsystem for Linux" (via bash). On Ubuntu on Windows 10, after creating the symlink like above, to add the folder to your $PATH add PATH="/usr/local/node/bin:$PATH" in the ~/.bashrc file.
I combined both of these answers for my docker container. I wanted the executable to be in the PATH already without me explicitly doing that.
#!/usr/bin/env bash
install_node() {
NODE_VERSION="v8.3.0"
curl -# "http://nodejs.org/dist/${NODE_VERSION}/node-${NODE_VERSION}-linux-x64.tar.gz" | tar -xz
cp -pr "node-${NODE_VERSION}-linux-x64" "/usr/local/"
ln -s "/usr/local/node-${NODE_VERSION}-linux-x64" "/usr/local/node"
ln -s /usr/local/node/bin/* "/usr/local/bin"
rm -rf "node-${NODE_VERSION}-linux-x64"
}
install_node
You can extract the binary anywhere and use update-alternatives command which maintain symbolic links determining default commands for example this is on my laptop.
first i extract my node node-v10.16.3-linux-x64.tar.xz on /mnt/e/WSL_Ubuntu/Downloads/node-v10.16.3-linux-x64/ folder :
xxxx#xxxxPC:.../WSL_Ubuntu/Downloads$ tar xvf node-v10.16.3-linux-x64.tar.xz
xxxx#xxxxPC:.../WSL_Ubuntu/Downloads$ cd node-v10.16.3-linux-x64/
then update-alternatives --install :
xxxx#xxxxPC:.../node-v10.16.3-linux-x64/bin$ sudo update-alternatives --install /home/wira/.local/bin/node node\
> /mnt/e/WSL_Ubuntu/Downloads/node-v10.16.3-linux-x64/bin/node 60
update-alternatives: using /mnt/e/WSL_Ubuntu/Downloads/node-v10.16.3-linux-x64/bin/node to provide /home/wira/.local/bin/node (node) in auto mode
Now i use the node on terminal
xxxx#xxxxPC:.../node-v10.16.3-linux-x64/bin$ node --version
v10.16.3
You should also update-alternatives --install on npm binaries.
I think there is still a cleaner way
NODE_VERSION="v6.7.0"
# Download
curl -LO http://nodejs.org/dist/$NODE_VERSION/node-$NODE_VERSION-linux-x64.tar.gz
# uncompress
tar xzf node-$NODE_VERSION-linux-x64.tar.gz
# selective copy
cp -R ./node-$NODE_VERSION-linux-x64/bin/* /usr/local/bin
cp -R ./node-$NODE_VERSION-linux-x64/lib/* /usr/local/lib
cp -R ./node-$NODE_VERSION-linux-x64/include/* /usr/local/include
cp -R ./node-$NODE_VERSION-linux-x64/share/* /usr/local/share
Node should be working now
$ node -e 'console.log("HI")'
HI
Hope it helps