When I use my new shared library I cannot access environment variables for any src class which is executed either directly by the Jenkinsfile or via a var/*.groovy script. This problem persists even when I add withEnv to the var/*groovy script.
What is the trick to get environment variables to propagate to jenkins shared library src class execution?
Jenkinsfile
withEnv(["FOO=BAR2"]) {
println "Jenkinsfile FOO=${FOO}"
library 'my-shared-jenkins-library'
lib.displayEnv()
Shared Library var/lib.groovy
def displayEnv() {
println "Shared lib var/lib FOO=${FOO}"
MyClass c = new MyClass()
}
Shared Library src/MyClass.groovy
class MyClass() {
MyClass() {
throw new Exception("Shared lib src/MyClass FOO=${System.getenv('FOO')")
}
}
** Run Result **
Jenkinsfile FOO=BAR
Shared lib var/lib FOO=BAR
java.lang.Exception: Shared lib src/MyClass FOO=null
...
It sure looks like the only way to handle this is to pass the this from Jenkins file down to the var/lib.groovy and harvest from that object
Jenkinsfile
withEnv(["FOO=BAR2"]) {
library 'my-shared-jenkins-library'
lib.displayEnv(this)
var/lib.groovy
def displayEnv(script) {
println "Shared lib var/lib FOO=${FOO}"
MyClass c = new MyClass(script)
}
src class
MyClass(def script) {
throw new Exception("FOO=${script.env.FOO}")
}
I believe you can populate the environment variable as below, where shared library can access.
Jenkisfile
env.FOO="BAR2"
library 'my-shared-jenkins-library'
lib()
vars/lib.groovy
def call(){
echo ("FOO: ${FOO}")
echo ("FOO:"+env.FOO)
}
Another method is use the "steps" variable:
In Jenkinsfile
mypackages.myclass.mymethod(steps)
In src
class myclass implements Serializable {
void mymethod(steps) {
String myEnvVar = steps.sh(returnStdout: true, script: "env | grep 'myVar' | cut -f 2- -d '='")
}
}
Not sure what the experts will say about the solution but I was able to access the variables defined in my Jenkinsfile from the shared library using evaluate.
Jenkinsfile
myVar = "abc"
vars/test.groovy
String myVar = evaluate("myVar")
For me this just works.
Jenkinsfile:
@Library('jenkins-library') _
pipeline {
agent any
environment {
FOO = 'bar'
}
stages {
stage('Build') {
steps {
script {
buildImage()
...
The library vars/buildImage.groovy:
def call() {
println(this.env.FOO)
println(env.FOO)
}
So to pass the environment to a class in the library, just use this
in the vars/yourfunc.groovy
.
I stumbled upon this problem lately, so I'm gonna add my $0.02
.
The basic template I use for var/*.groovy is:
// var/myMethod.groovy
import cool.package.Clazz
def call(Map m) {
m.put('env', env)
m.put('steps', steps)
new Clazz(m).call()
}
And the template for src/**/*.groovy
// src/cool/package/Clazz.groovy
class Clazz {
private String cool_field_1 = "default-value-1"
private int cool_value = 42
def env
def steps
def call() {
steps.echo("env.BUILD_TAG: ${env.BUILD_TAG}")
//...
}
}
In Jenkinsfile
it is used standard way:
@Library('mylib@mybranch')
pipeline {
stages {
stage('St 1') {
steps { myMethod('cool_value': 43) }
}
}
}
Disclaimer: I don't do Groovy but since it looks similar to Java, so I can use it little. Also using Map
seems to give the advantage of quite flexible interface.
来源:https://stackoverflow.com/questions/48234633/access-environment-variables-in-jenkins-shared-library-code