Jenkins Pipeline - set and use environment variables
How to list the environment variables available to Jenkins Pipeline
examples/jenkins/list_environment.Jenkinsfile
pipeline {
agent none
environment {
color = "blue"
}
stages {
stage('first') {
agent { label 'master' }
steps {
sh "printenv | sort"
}
}
}
}
In this example we list the environment variables using the printenv command of Unix/Linux which we pipe through the Unix sort command so we'll see the environment variables in a sorted list.
We invoke it using the sh command of the Jenkins Pipeline.
Before we do that we set a new variable called "color" in the environment section of the Jenkins Pipeline.
On Unix/Linux:
sh('printenv | sort')
On Windows you could run:
bat('set')
The output will looks something like this:
BUILD_DISPLAY_NAME=#18 BUILD_ID=18 BUILD_NUMBER=18 BUILD_TAG=jenkins-list_environment_variables-18 BUILD_URL=http://localhost:8080/job/list_environment_variables/18/ DBUS_SESSION_BUS_ADDRESS=unix:path=/run/user/110/bus EXECUTOR_NUMBER=1 HOME=/var/lib/jenkins HUDSON_COOKIE=40cd788e-91bd-4ebd-86c9-c10333fa27a9 HUDSON_HOME=/var/lib/jenkins HUDSON_SERVER_COOKIE=912830efeb6e2316 HUDSON_URL=http://localhost:8080/ JENKINS_HOME=/var/lib/jenkins JENKINS_NODE_COOKIE=dbf878e6-0ae5-4ffe-a32c-aa7876f975ce JENKINS_SERVER_COOKIE=durable-f6e3ca8e5d2310d4d5695d128db1ea2f JENKINS_URL=http://localhost:8080/ JOB_BASE_NAME=list_environment_variables JOB_DISPLAY_URL=http://localhost:8080/job/list_environment_variables/display/redirect JOB_NAME=list_environment_variables JOB_URL=http://localhost:8080/job/list_environment_variables/ LANG=C.UTF-8 LOGNAME=jenkins MAIL=/var/mail/jenkins NODE_LABELS=master NODE_NAME=master PATH=/usr/local/bin:/usr/bin:/bin:/usr/local/games:/usr/games:/snap/bin PWD=/var/lib/jenkins/workspace/list_environment_variables RUN_CHANGES_DISPLAY_URL=http://localhost:8080/job/list_environment_variables/18/display/redirect?page=changes RUN_DISPLAY_URL=http://localhost:8080/job/list_environment_variables/18/display/redirect SHELL=/bin/bash SHLVL=1 STAGE_NAME=example USER=jenkins WORKSPACE=/var/lib/jenkins/workspace/list_environment_variables XDG_DATA_DIRS=/usr/local/share:/usr/share:/var/lib/snapd/desktop XDG_RUNTIME_DIR=/run/user/110 XDG_SESSION_ID=c1 _=/usr/bin/daemon color=blue
List the environment variables without running the shell
The env object is of type class org.jenkinsci.plugins.workflow.cps.EnvActionImpl.
examples/jenkins/list_environment_internally.Jenkinsfile
pipeline {
agent none
environment {
color = 'blue'
}
stages {
stage('example') {
agent { label 'master' }
steps {
script {
def fields = env.getEnvironment()
fields.each {
key, value -> println("${key} = ${value}");
}
println(env.PATH)
}
}
}
}
}
We use the getEnvironment method that returns a hudson.EnvVars object. This is a Groovy map so we can already use the each method to go over the keys and values.
This however only seem to list the internal variables of Jenkins and for example PATH was not in the list even though we can access it as env.PATH.
How to set environment variables in Jenkins?
As also seen above, a section called environment can be added to the pipeline
environment {
SOME_NAME = "some value"
}
Then it can be accessed using the following syntax:
env.SOME_NAME
pipeline {
agent none
environment {
field = 'some'
}
stages {
stage ('Preparation') {
agent { label 'master'}
environment {
JENKINS_PATH = sh(script: 'pwd', , returnStdout: true).trim()
}
steps {
echo "Hello world"
echo "PATH=${JENKINS_PATH}"
sh 'echo "JP=$JENKINS_PATH"'
}
}
}
}
How to use environment variables in the environment section of Jenkins?
environment {
PATH = "/path/to/dir:${env.PATH}"
JNK_PATH = "${env.WORKSPACE}\\subdir"
}
The above only works when the environment section is inside a "stage" but would yield "null" for WORKSPACE outside the stages.
Set environment executing code
environment {
field = func()
}
def func() {
...
return "value"
}
A working example:
examples/jenkins/set_environment_by_script.Jenkinsfile
pipeline {
agent none
environment {
first_path = get_first()
}
stages {
stage('example') {
agent { label 'master' }
steps {
print(env.first_path)
}
}
}
}
def get_first() {
node('master') {
return env.PATH.split(':')[0]
}
}
Comments
I am trying to do something like this and getting error. Can I use conditional assignments in the environment section?
environment{ if (env == "e2e") { SLACK_CHANNEL = "#e2e-monitoring" } else if (env == "pqa") { SLACK_CHANNEL = "#pqa-monitoring" } }
so beautiful article!!!!
Thanks for this tutorial. However, I'm having hard times understanding how to pass environment variables in case of Jenkins Pipelines by NOT using Jenkinsfile. In case of simple Jenkins build projects it was easy to do. However, how can I add custom environment variables to Jenkins Pipeline Multibranch build project so that the variables are not in Jenkinsfile?
I don't want to add sensitive keys, passwords etc to Jenkinsfile that would go to version control system.
Any tips?
---
You can add "secrets" to your Jenkins server and then those secrets will be available to your jobs, but won't be part of your Jenkinsfile. https://www.jenkins.io/doc/book/using/using-credentials/
---
Thanks! That's what I figured out, too but it's very cumbersome way of doing this:( Right now I went back to old way of creating builds - step by step and I do stuff in "Shell" step where I can set env variables easily.
Strange that some developers came up with that idea that when you add env variables in your pipeline steps these will be written and committed back to your git repo. That's VERY dangerous. Just wondering :)
---
Why would you add the environment variables to your repo? The link I shared does not do that.
---
I would not and as I explained - it's very bad idea. I tried to explain that the standard way of doing it in the pipelines (add env variables by each step or to the pipeline) is designed in a really bad way so that whatever changes you do to your pipeline in Jenkins - it would get committed to SCM.
I added screenshot to illustrate my problem.
---
I did not see that happening.
i'm new in jenkins, i would know if someone know how can i build a remote trigger in the multibranch pipeline with post commit hook ? and how can i get the current name branch in Pipeline job because i want to do some stages in master branch and other in develop branch for example when i try with echo env.BRANCH_NAME it's return NULL :(
can I make all the environment variables as None before my pipeline starts? and then set few values if I want. Is it possible?
--- I am not sure I understand you. Why would you want to do that? What are you trying to achieve?
The environment variables come from the system, without them probably Jenkins would not be able to run. However you can set the ones you like in your Jenkins file first to be none and then to some interesting value.
--- I am using these env.vars to set few jenkins parameters in my groovy. Because of which they are retaining previous builds parameters. I am trying to find a way to not retain the previous build's values. --- Jenkins gives you access to some meta data of the previous builds, but you don't need to access them.
What environment variables do you see that come from an earlier build?
---
I am not using any system env variables to set to the parameters. Am setting some parameters to my jenkins job, by doing env.build="yyyyyyy" and am setting the parameters to env.build as default values if its not set in the current build and is none. I see even though am not setting any env.build next time, its value is being taken from the previous build and is set to the jenkjns parameter.
---
Could you create a minimal example that shows this?
---
def string_param(Map opt=[defaultValue:"", description:""]) {
if (opt.name == null) { throw new Exception("Arg name is mandatory") }
return string(name: "$opt.name", defaultValue: "$opt.defaultValue", description: "$opt.description") }
def def_boards = env."${board_u}_BOARDS" def boards = string_param(name: "${board}_boards", defaultValue: "${def_boards?:''}", description: "Enter $board boards example: abc-1,abc-2")
in the first build when I set, env.xyz_BOARDS="xxxxxxxxxxxxx", for the second build even though am not setting it, this value is being retailed to boards value. How can I stop that for being retained?
---
It is unclear to me why Jenkins keeps the old values, but at the end of your job you could set them to null again to make sure they are clean.
---
Yeah thank you. I tried resetting the vars in the start of the groovy so that it would work for concurrent builds as well and I see its working as intended. Thank you
Can you make a article on Jenkins shared libs ?
env.first_path
calling a method that way is not something I would want to do too often, it just looks like a field, and it may/can have side-effects. Good explainer of the gotchas , thanks for sharing :-)
Published on 2019-02-15