I have tried a number of different ways to assign an array inside a RUN command within a Dockerfile. None of them seem to work. I am running on Ubuntu-Slim, with bash as my default shell.
I've tried this (second line below)
RUN addgroup --gid 1000 node \
&& NODE_BUILD_PACKAGES=("binutils-gold" "g++" "gcc" "gnupg" "libgcc-7-dev" "linux-headers-generic" "make" "python3" ) \
...
But it fails with /bin/sh: 1: Syntax error: "(" unexpected.
I also tried assigning it as an ENV variable, as in:
ENV NODE_BUILD_PACKAGES=("binutils-gold" "g++" "gcc" "gnupg" "libgcc-7-dev" "linux-headers-generic" "make" "python3" )
but that fails as well.
Assigning and using arrays in Bash is completely supported. Yet, it appears that I can't use that feature of Bash when running in a Dockerfile. Can someone confirm/deny that you can assign arrays variables inside of shell commands in Dockerfile RUN syntax (or ENV variable syntax)?
The POSIX shell specification does not have arrays. Even if you're using non-standard shells like GNU bash, environment variables are always simple strings and never hold arrays either.
The default shell in Docker is usually /bin/sh, which should conform to the POSIX spec, and not bash. Alpine-based images don't have bash at all unless you go out of your way to install it. I'd generally recommend trying to stick to the POSIX syntax whenever possible.
A typical Dockerfile is fairly straightforward; it doesn't have a lot of parts that get reused multiple times and most of the things you can specify in a Dockerfile you don't need to be user-configurable. So for a list of OS packages, for example, I'd just list them out in a RUN command and not bother trying to package them into a variable.
RUN apt-get update \
&& DEBIAN_FRONTEND=noninteractive \
apt-get install --no-install-recommends --assume-yes \
binutils-gold \
g++ \
gcc \
...
Other things I see in Stack Overflow questions that do not need to be parameterized include the container path (set it once as the WORKDIR and refer to . thereafter), the process's port (needs to be a fixed number for the second docker run -p part), and user IDs (can be overridden with docker run -u, and you don't usually want to build an image that can only run on one system).
WORKDIR /app # not an ENV or ARG
COPY . . # into the WORKDIR, do not need to repeat
RUN adduser node # with no specific uid
EXPOSE 3000 # a fixed port number
RUN mkdir /data # also use a fixed path for potential mount points
You can have array NODE_BUILD_PACKAGES in RUN if you define SHELL :
SHELL ["/bin/bash", "-c"]
Running pods have some environment variables defined inside, for example:
/ # printenv
REACT_APP_ENV_VARIABLE=Variable from Kube!
REDIS_SERVICE_PORT=6379
KUBERNETES_SERVICE_PORT=443
KUBERNETES_PORT=tcp://10.38.0.1:443
REDIS_PORT=tcp://10.38.61.225:6379
REDIS_PORT_6379_TCP_ADDR=10.38.61.225
HOSTNAME=playground-pod
PLAYGROUND_SERVICE_SERVICE_HOST=10.38.0.53
REDIS_PORT_6379_TCP=tcp://10.38.61.225:6379
PLAYGROUND_SERVICE_SERVICE_PORT=80
PLAYGROUND_SERVICE_PORT=tcp://10.38.0.53:80
PLAYGROUND_SERVICE_PORT_80_TCP_ADDR=10.38.0.53
KUBERNETES_PORT_443_TCP_PROTO=tcp
PLAYGROUND_SERVICE_PORT_80_TCP_PORT=80
PLAYGROUND_SERVICE_PORT_80_TCP_PROTO=tcp
REACT_APP_ENV_VARIABLE_TWO=192.168.1.12
PLAYGROUND_SERVICE_PORT_80_TCP=tcp://10.38.0.53:80
How should I configure a React app like this one:
function App() {
return (
<div className="App">
<header className="App-header">
<p>
<code>ENV. VARIABLE: </code> {x.REACT_APP_ENV_VARIABLE}
</p>
</header>
</div>
);
}
export default App;
to read and inject some of the variables present in the pod?
The main reason I want to know it, is dynamic update of e.g. backend or Redis URLs - they might change when the app is restarted, rescheduled, etc.
My first approach was using a config.json file imported to the app, but this way I can't import dynamic values generated by running pods.
You can use the library dot-env
import React from "react";
import env from "react-dotenv";
export function MyComponent() {
return <div>{env.REACT_APP}</div>;
}
while in deployment you can pass secret from secret and configmap
spec:
containers:
- name: example-site
image: example/app:v1
ports:
- containerPort: 80
env:
- name: REACT_APP
value: "123456"
The main reason I want to know it, is dynamic update of e.g. backend
or Redis URLs - they might change when the app is restarted,
rescheduled, etc.
Above scenario fit well with your requirement, instead of using the config.json.
You can pass multiple values to deployment using configmap and secrets.
As #Harsh Manvar suggested (thanks a lot!), the react-dotenv library can be used, but just adding it to the project is not enough.
Firstly, you have to follow all steps described in react-dotenv documentation (adding .env file to your project, editing package.json file).
In my case, .env file looked like this:
REACT_APP_DEPLOY_SETUP='__dps__'
REACT_APP_PORT='__prt__'
REACT_APP_BACKEND_URL='__bur__'
These are just placeholders for real values that will be added during runtime.
Having the .env file ready, npm scripts prepended with react-dotenv command, and the variables whitelisted (as described in the library documentation), you can build your app image.
When the image is ready, add it to the kubernetes pod config file and replace your variables' placeholders with real values, like this:
[...]
spec:
containers:
- name: plg-frontend
image: localhost:5000/frontend:1.22
ports:
- containerPort: 80
command:
- sh
- -c
args:
- sed -i "s/__prt__/$REACT_APP_PORT/g" /usr/share/nginx/html/env.js;
sed -i "s/__bur__/http:\/\/$BACKEND_SERVICE_HOST/g" /usr/share/nginx/html/env.js;
sed -i "s/__dps__/$REACT_APP_DEPLOY_SETUP/g" /usr/share/nginx/html/env.js;
nginx -g 'daemon off;'
env:
- name: REACT_APP_DEPLOY_SETUP
value: "development"
- name: REACT_APP_PORT
value: "5089"
What happened up there, was replacing placeholder values with actual values:
$BACKEND_SERVICE_HOST is an environment variable that exists in the pod and can be read from the running container,
$REACT_APP_DEPLOY_SETUP is a regular string defined by user,
$REACT_APP_PORT is an integer value (it has to be in quotes, like strings!).
And the replacing happened with sed command (or rather: sh -c "sed -i ..."). All of the commands are chained, so don't forget about semicolons at the end of each argument.
All of the replacements were made in /usr/share/nginx/html/env.js file, which is created by react-dotenv library in project root. The actual location depends on where you mounted your build image (it's defined in Dockerfile).
Lastly, nginx command is called, since this is the final command invoked in image's Dockerfile. Without this explicit call, the command from the Dockerfile would be overwritten with the commands related to the pod container and, in this case, nginx wouldn't start your app.
After the pod is started, you can check whether the variables are present in the container:
kubectl exec <pod-name> -- printenv | grep REACT_APP
But it doesn't mean they were read by your app during runtime. To see if they were changed to the values from the pod definition, you can either exec running pod container and preview the env.js file or add some console logging in the app code.
I'm trying to write a small script to change the current directory to my project directory:
#!/bin/bash
cd /home/tree/projects/java
I saved this file as proj, added execute permission with chmod, and copied it to /usr/bin. When I call it by:
proj, it does nothing. What am I doing wrong?
Shell scripts are run inside a subshell, and each subshell has its own concept of what the current directory is. The cd succeeds, but as soon as the subshell exits, you're back in the interactive shell and nothing ever changed there.
One way to get around this is to use an alias instead:
alias proj="cd /home/tree/projects/java"
You're doing nothing wrong! You've changed the directory, but only within the subshell that runs the script.
You can run the script in your current process with the "dot" command:
. proj
But I'd prefer Greg's suggestion to use an alias in this simple case.
The cd in your script technically worked as it changed the directory of the shell that ran the script, but that was a separate process forked from your interactive shell.
A Posix-compatible way to solve this problem is to define a shell procedure rather than a shell-invoked command script.
jhome () {
cd /home/tree/projects/java
}
You can just type this in or put it in one of the various shell startup files.
The cd is done within the script's shell. When the script ends, that shell exits, and then you are left in the directory you were. "Source" the script, don't run it. Instead of:
./myscript.sh
do
. ./myscript.sh
(Notice the dot and space before the script name.)
To make a bash script that will cd to a select directory :
Create the script file
#!/bin/sh
# file : /scripts/cdjava
#
cd /home/askgelal/projects/java
Then create an alias in your startup file.
#!/bin/sh
# file /scripts/mastercode.sh
#
alias cdjava='. /scripts/cdjava'
I created a startup file where I dump all my aliases and custom functions.
Then I source this file into my .bashrc to have it set on each boot.
For example, create a master aliases/functions file: /scripts/mastercode.sh
(Put the alias in this file.)
Then at the end of your .bashrc file:
source /scripts/mastercode.sh
Now its easy to cd to your java directory, just type cdjava and you are there.
You can use . to execute a script in the current shell environment:
. script_name
or alternatively, its more readable but shell specific alias source:
source script_name
This avoids the subshell, and allows any variables or builtins (including cd) to affect the current shell instead.
Jeremy Ruten's idea of using a symlink triggered a thought that hasn't crossed any other answer. Use:
CDPATH=:$HOME/projects
The leading colon is important; it means that if there is a directory 'dir' in the current directory, then 'cd dir' will change to that, rather than hopping off somewhere else. With the value set as shown, you can do:
cd java
and, if there is no sub-directory called java in the current directory, then it will take you directly to $HOME/projects/java - no aliases, no scripts, no dubious execs or dot commands.
My $HOME is /Users/jleffler; my $CDPATH is:
:/Users/jleffler:/Users/jleffler/mail:/Users/jleffler/src:/Users/jleffler/src/perl:/Users/jleffler/src/sqltools:/Users/jleffler/lib:/Users/jleffler/doc:/Users/jleffler/work
Use exec bash at the end
A bash script operates on its current environment or on that of its
children, but never on its parent environment.
However, this question often gets asked because one wants to be left at a (new) bash prompt in a certain directory after execution of a bash script from within another directory.
If this is the case, simply execute a child bash instance at the end of the script:
#!/usr/bin/env bash
cd /home/tree/projects/java
echo -e '\nHit [Ctrl]+[D] to exit this child shell.'
exec bash
To return to the previous, parental bash instance, use Ctrl+D.
Update
At least with newer versions of bash, the exec on the last line is no longer required. Furthermore, the script could be made to work with whatever preferred shell by using the $SHELL environment variable. This then gives:
#!/usr/bin/env bash
cd desired/directory
echo -e '\nHit [Ctrl]+[D] to exit this child shell.'
$SHELL
I got my code to work by using. <your file name>
./<your file name> dose not work because it doesn't change your directory in the terminal it just changes the directory specific to that script.
Here is my program
#!/bin/bash
echo "Taking you to eclipse's workspace."
cd /Developer/Java/workspace
Here is my terminal
nova:~ Kael$
nova:~ Kael$ . workspace.sh
Taking you to eclipe's workspace.
nova:workspace Kael$
simply run:
cd /home/xxx/yyy && command_you_want
When you fire a shell script, it runs a new instance of that shell (/bin/bash). Thus, your script just fires up a shell, changes the directory and exits. Put another way, cd (and other such commands) within a shell script do not affect nor have access to the shell from which they were launched.
You can do following:
#!/bin/bash
cd /your/project/directory
# start another shell and replacing the current
exec /bin/bash
EDIT: This could be 'dotted' as well, to prevent creation of subsequent shells.
Example:
. ./previous_script (with or without the first line)
On my particular case i needed too many times to change for the same directory.
So on my .bashrc (I use ubuntu) i've added the
1 -
$ nano ~./bashrc
function switchp
{
cd /home/tree/projects/$1
}
2-
$ source ~/.bashrc
3 -
$ switchp java
Directly it will do: cd /home/tree/projects/java
Hope that helps!
It only changes the directory for the script itself, while your current directory stays the same.
You might want to use a symbolic link instead. It allows you to make a "shortcut" to a file or directory, so you'd only have to type something like cd my-project.
You can combine Adam & Greg's alias and dot approaches to make something that can be more dynamic—
alias project=". project"
Now running the project alias will execute the project script in the current shell as opposed to the subshell.
You can combine an alias and a script,
alias proj="cd \`/usr/bin/proj !*\`"
provided that the script echos the destination path. Note that those are backticks surrounding the script name.
For example, your script could be
#!/bin/bash
echo /home/askgelal/projects/java/$1
The advantage with this technique is that the script could take any number of command line parameters and emit different destinations calculated by possibly complex logic.
to navigate directories quicky, there's $CDPATH, cdargs, and ways to generate aliases automatically
http://jackndempsey.blogspot.com/2008/07/cdargs.html
http://muness.blogspot.com/2008/06/lazy-bash-cd-aliaes.html
https://web.archive.org/web/1/http://articles.techrepublic%2ecom%2ecom/5100-10878_11-5827311.html
In your ~/.bash_profile file. add the next function
move_me() {
cd ~/path/to/dest
}
Restart terminal and you can type
move_me
and you will be moved to the destination folder.
You can use the operator && :
cd myDirectory && ls
While sourcing the script you want to run is one solution, you should be aware that this script then can directly modify the environment of your current shell. Also it is not possible to pass arguments anymore.
Another way to do, is to implement your script as a function in bash.
function cdbm() {
cd whereever_you_want_to_go
echo "Arguments to the functions were $1, $2, ..."
}
This technique is used by autojump: http://github.com/joelthelion/autojump/wiki to provide you with learning shell directory bookmarks.
You can create a function like below in your .bash_profile and it will work smoothly.
The following function takes an optional parameter which is a project.
For example, you can just run
cdproj
or
cdproj project_name
Here is the function definition.
cdproj(){
dir=/Users/yourname/projects
if [ "$1" ]; then
cd "${dir}/${1}"
else
cd "${dir}"
fi
}
Dont forget to source your .bash_profile
This should do what you want. Change to the directory of interest (from within the script), and then spawn a new bash shell.
#!/bin/bash
# saved as mov_dir.sh
cd ~/mt/v3/rt_linux-rt-tools/
bash
If you run this, it will take you to the directory of interest and when you exit it it will bring you back to the original place.
root#intel-corei7-64:~# ./mov_dir.sh
root#intel-corei7-64:~/mt/v3/rt_linux-rt-tools# exit
root#intel-corei7-64:~#
This will even take you to back to your original directory when you exit (CTRL+d)
I did the following:
create a file called case
paste the following in the file:
#!/bin/sh
cd /home/"$1"
save it and then:
chmod +x case
I also created an alias in my .bashrc:
alias disk='cd /home/; . case'
now when I type:
case 12345
essentially I am typing:
cd /home/12345
You can type any folder after 'case':
case 12
case 15
case 17
which is like typing:
cd /home/12
cd /home/15
cd /home/17
respectively
In my case the path is much longer - these guys summed it up with the ~ info earlier.
As explained on the other answers, you have changed the directory, but only within the sub-shell that runs the script. this does not impact the parent shell.
One solution is to use bash functions instead of a bash script (sh); by placing your bash script code into a function. That makes the function available as a command and then, this will be executed without a child process and thus any cd command will impact the caller shell.
Bash functions :
One feature of the bash profile is to store custom functions that can be run in the terminal or in bash scripts the same way you run application/commands this also could be used as a shortcut for long commands.
To make your function efficient system widely you will need to copy your function at the end of several files
/home/user/.bashrc
/home/user/.bash_profile
/root/.bashrc
/root/.bash_profile
You can sudo kwrite /home/user/.bashrc /home/user/.bash_profile /root/.bashrc /root/.bash_profile to edit/create those files quickly
Howto :
Copy your bash script code inside a new function at the end of your bash's profile file and restart your terminal, you can then run cdd or whatever the function you wrote.
Script Example
Making shortcut to cd .. with cdd
cdd() {
cd ..
}
ls shortcut
ll() {
ls -l -h
}
ls shortcut
lll() {
ls -l -h -a
}
If you are using fish as your shell, the best solution is to create a function. As an example, given the original question, you could copy the 4 lines below and paste them into your fish command line:
function proj
cd /home/tree/projects/java
end
funcsave proj
This will create the function and save it for use later. If your project changes, just repeat the process using the new path.
If you prefer, you can manually add the function file by doing the following:
nano ~/.config/fish/functions/proj.fish
and enter the text:
function proj
cd /home/tree/projects/java
end
and finally press ctrl+x to exit and y followed by return to save your changes.
(NOTE: the first method of using funcsave creates the proj.fish file for you).
You need no script, only set the correct option and create an environment variable.
shopt -s cdable_vars
in your ~/.bashrc allows to cd to the content of environment variables.
Create such an environment variable:
export myjava="/home/tree/projects/java"
and you can use:
cd myjava
Other alternatives.
Note the discussion How do I set the working directory of the parent process?
It contains some hackish answers, e.g.
https://stackoverflow.com/a/2375174/755804 (changing the parent process directory via gdb, don't do this) and https://stackoverflow.com/a/51985735/755804 (the command tailcd that injects cd dirname to the input stream of the parent process; well, ideally it should be a part of bash rather than a hack)
It is an old question, but I am really surprised I don't see this trick here
Instead of using cd you can use
export PWD=the/path/you/want
No need to create subshells or use aliases.
Note that it is your responsibility to make sure the/path/you/want exists.
I have to work in tcsh, and I know this is not an elegant solution, but for example, if I had to change folders to a path where one word is different, the whole thing can be done in the alias
a alias_name 'set a = `pwd`; set b = `echo $a | replace "Trees" "Tests"` ; cd $b'
If the path is always fixed, the just
a alias_name2 'cd path/you/always/need'
should work
In the line above, the new folder path is set
This combines the answer by Serge with an unrelated answer by David. It changes the directory, and then instead of forcing a bash shell, it launches the user's default shell. It however requires both getent and /etc/passwd to detect the default shell.
#!/usr/bin/env bash
cd desired/directory
USER_SHELL=$(getent passwd <USER> | cut -d : -f 7)
$USER_SHELL
Of course this still has the same deficiency of creating a nested shell.