I have a little node application on a server (node mailer) that I run by going to its source folder and executing npm start. I figured the best way to run this automatically would be to create a my_script.sh file and drop it in the init.d directory of my debian box. Inside the file (below the !#/bin/bash line), the code to execute is
'/opt/mycode/source/npm start'
I save the line to the .sh file and restarted the machine, but so far haven't got it to work. My question is: is this even how you start a script like this (using that command and an .sh file)? It does start normally when I do it manually (when I navigate to it and run npm start in the terminal). I included the single quotes around it because of the space between npm start. Also, if I want to verify that it worked, which process would I look for other than just pinging my smtp mailer? Finally, I know I need to run:
update-rc.d my_script.sh defaults
but I was also confused at to whether I had done this correctly either (is it just the name of the file that goes there or the file plus the extension)?
The script that you leave on the init.d folder should not have any extension and should have functions to start, stop and get the status of the service (your application).
I'll leave a link with an example as well as with some basis in order to build the Linux service script.
I would suggest reloading the daemon with systemctl daemon-reload in order to refresh the Linux service files once you add a new one.
I've installed nodejs as described here.
Everything works fine when I ssh to the server myself. But I've created a script that deploys my application and call it via bitbucket pipelines. Everything else works fine (composer install, php artisan migrate etc.), except npm install. The error message is
./deploy: line 26: npm: command not found
In bitbucket-pipelines.yml I call my script like this:
- step:
script:
- ssh user#ip_address ./deploy
When I call the script by myself everything works. The problem is only with bitbucket pipelines. I have no clue why this happens :(.
Running which composer revealed that at least composer command is not getting picked up from your assumed location i.e., ~/composer dir. Instead, it was found in /opt/cpanel/composer/bin/composer.
which npm returned the following:
no npm in (/usr/local/cpanel/3rdparty/lib/path-bin:/usr/local/jdk/bin:/usr/kerberos/sbin:/usr/kerberos/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin:/usr/X11R6/bin:/usr/local/bin:/usr/X11R6/bin:/root/bin:/opt/bin:/opt/cpanel/composer/bin:/home/handmast/.local/bin:/home/handmast/bin)
Now since you are able to manually run the command npm, you just need to figure about the path from where npm is running and ensure that the path is explicitly added to the user's ~/.bashrc file and things should work fine. You need to do this because as per your observation, ssh is not able to find it.
export PATH=$PATH:/path/to/npm/binary
Note: To avoid any confusion, just remember that while adding the path to your binary, you just have to add the path to the directory where npm resides. Don't add npm at the end. For example, following is incorrect way:
export PATH=$PATH:/home/handmast/bin/npm
Correct way is this:
export PATH=$PATH:/home/handmast/bin
Before I start, I want to say that I already checked these answers:
Jenkins build step fails on 'npm install <whatever>'
Jenkin's build failing on npm install
Now, I'm dealing with this issue for a while already and thus I tried a bunch of stuff.
Firstly, I installed node + npm via homebrew. A simple $ node -v and $ npm -v echoed the version v0.10.36 for node and v2.3.* for npm, which also means I HAVE THEM IN THE PATH and they work while called in the terminal.
Simply adding node -v; npm -v to the execute shell in Jenkins didn't do it. After a bit of tinkering I copied what $: which node yielded in the terminal to the above mentioned script, which now looked like this: /usr/local/bin/node and apparently that worked. The Jenkins build succeeded and 'node-v0.10.36' was proudly displayed in the console output.
When doing the same for 'npm' which happened to be /usr/local/bin/npm --version the computing gods weren't so merciful anymore. A big 'env: node: No such file or directory' error was thrown this time and the whole build failed.
The actual command that fails is
$ /bin/sh -xe /var/folders/wr/g_dl81tn5_x0t_yz3jw602cr0000gn/T/hudson8770480548136671253.sh and "surprisingly" when I run the same command in the terminal it succeeds.
I also uninstalled the homebrew node & npm versions and installed them afterwards via the package manager. Same results.
Ultimately I also did this: https://gist.github.com/DanHerbert/9520689, with no luck.
Notes:
I'm running Jenkins 1.613 and tried with 1.5**
I didn't create a "Jenkins" specific user but instead I'm using the admin. This happens to be the same user that Jenkins runs, since the who am i command inside the executable script yields the admin's user name.
sudo'ing doens't help
I'm also running the whole thing in a Virtual Environment - vagrant
I'm not running Jenkins as a deamon, as it's conflicting with xtools, but as a simple process
I also tried out jenkins-node plugin with various configs (can detail if needed)
Thanks a lot for your help, and let me know if you need any other info, screenshots, logs, etc.
I found my own solution. The problem was that the PATH although visible in shell was not exported for the Jenkins job, and so, the first workaround, as found here, was to export it in the actual script like so:
but this feels like a hack!
The right and elegant solution is to use Jenkins EnvInject Plugin and export the path in the added Properties content textarea on the configuration page, like so:
Manage Jenkins -> Configure System -> Global properties -> Environment variables
I have installed node on my machine and have two executables, node.exe and nodejs.exe in my directory. Through the NPM I have installed grunt and jshint for grunt; following this guide:
http://strongloop.com/strongblog/use-grunt-js-and-the-power-of-javascript-to-automating-repetitive-tasks/
When I try and run "grunt" "node" or "grunt jshint" into the windows command prompt or Git Bash then I recieve the following response:
The node identifier for <machine number> is <id number>
I have in my path the location to the \nodejs\node.exe file
I do not have admin or root access to my machine, but can put requests in to add permissions or extra items to the Path. However, I was able to get JSHint to work in Sublime Text by specifying the node path as "nodejs/nodejs.exe"
Would changing my path to nodejs.exe instead fix this; or is there a deeper issue into the setup I have configured that would be causing this problem?
check your system %PATH% variable (echo %PATH%). Most certainly there is another application called node.bat, node.cmd or node.exe that is in your PATH before nodejs and thats why the system finds this first. I'm 99% sure that this is not the output from nodejs
if you have access to your path try to prepend the nodejs path
I'm having a problem running Grunt from the command line on my Windows 8 machine.
My research indicates the most common solution is to install grunt-cli, since Grunt is no longer global. I also need to make sure I actually install the Grunt task runner, since that's not installed with grunt-cli.
Other solutions point to the PATH system environment variable, but that appears to be pointed as I'd expect to:
C:\Users[username]\AppData\Roaming\npm
Having done all that, I'm still getting a "'grunt' is not recognized as an internal or external command, operable program or batch file" error message in the CLI. I've tried the following things, uninstalling everything after every attempt:
Installed grunt-cli globally (npm install -g grunt-cli), then grunt at the directory level I want to use it (npm install grunt)
The same as above, but with the order of installation reversed
The same as both of the above, but using the Admin Command Prompt
Am I missing something obvious?
I've not had any issues with grunt on several different windows 8 machines.
If you open the folder: C:\Users\[username]\AppData\Roaming\npm
Do you have a file named grunt.cmd in this folder?
If not I'd maybe try npm install -g grunt-cli again, maybe from an elevated command prompt.
If this exists and you have C:\Users\[username]\AppData\Roaming\npm in your PATH environment variable then typing grunt from a command prompt should work.
Silly question, have you tried closing the command prompt and opening a new one?
Confirm your PATH is correct (and not messed up). Just type PATH from the command prompt. There's really no other explanation that makes sense given the error you're describing and the steps you've taken.
Normally, using the where grunt command would have found grunt.cmd in your path if npm is installed correctly and it has been properly added to the system path.
Close all Command Prompt instances.
Start a new Command Prompt instance.
Type PATH Enter and verify if C:\Users\Username\AppData\Roaming\npm is part of the path.
If not, you need to log off and on again,or close the Command Prompt and restart the explorer process.
In the Command Prompt, type where grunt Enter.
You're good if it reports:
C:\Users\Username\AppData\Roaming\npm\grunt
C:\Users\Username\AppData\Roaming\npm\grunt.cmd
Otherwise, you have to reinstall the grunt-cli package if it reports:
INFO: Could not find files for the given pattern(s).
Apparently, programs that change the PATH environment variable must broadcast a WM_SETTINGCHANGE message. The Windows' System settings window does it correctly when you change the PATH variable, but the NPM installer doesn't. That's why you have to restart explorer (or log off or restart, which has the same effect).
I know this has been answered but I thought I'd offer my step by step solution for windows 8.
First thing I checked was the PATH in my laptops Environment Variables (Right click my computer > properties > advanced system settings > Environment Variables)
It wasn't listed in there so I added a new variable in User variables (so it was specific only to my user account)
In the new user variable prompt I entered the following;
Variable Name: PATH
Variable Value: %USERPROFILE%\AppData\Roaming\npm
Quit command prompt, repoened, navigated to my projects directory and tried running grunt again and... SUCCESS!
I had the same issue.
I tried different things:
Restart computer
Deleted the grunt folder and ran
npm install -g grunt -cli
Didn't work.
Finally tried:
npm install -g grunt-cli
Worked perfectly.
Tried
where grunt
and I saw 2 locations where it was found.
I was facing the same problem on windows 8
I have added ' %APPDATA%\npm ' to the path variable . It has been working fine.
some times NPM install corrupts the basic windows path. i usually have a copy of my own version of PATH mainted separately. every week or on some installs i manually configure and update the %PATH% variable.
Basically Grunt.cmd is not availbe through %PATH% variable.
I have stucked with problem on Windows 8, that after install grunt-cli I've always got "command not found" while I'm tried to check grunt -v or where grunt. So I've added to enviroment PATH this path C:\Program Files (x86)\Git\local and run grunt.cmd from that folder (you need to look in node_modules folder here). And after reloading my terminal everything started to work.
Same happened to me and here was the solution:
Have you got 2 different versions of Node.JS installed?
Maybe Nodist?
This means you likely got NPM installed twice which will install the commands into 2 different folders:
Once into C:\Users\<user>\AppData\Roaming\npm and once into C:\dev\nodist\bin\bin.
C:\dev\nodist\bin\bin wasn't on my path variable so I added it, and I removed the Node.JS version I didn't want to use.
If you have no grunt.cmd file created by npm, make sure that you do not have a .npmrc in your home directory with: bin-links=false in it.
After getting a tonne of "'grunt' is not recognized as an internal or external command," errors, I solved this on Windows 10 by going to Path and adding C:\Users\Username\AppData\Roaming\npm