I am triggering pipeline2 from pipeline1.
Inside jenkinsfile of pipeline1 I have something like the following
build job: 'pipeline2'
But this doesn't post the live logs of pipeline2 in pipeline1. Is there a way to post the live logs of the downstream job in upstream job console?
This can partially be achieved by returning a build object,
buildobj = build job: 'pipeline2'
echo buildobj.rawBuild.log
But this doesn't echo live logs rather prints everything at once after build is executed.
Related
I have a pipeline that I want to run e2e test after deploying to uat environment, the e2e test runs on gitlab. What I want to do is to let gitlab use webhook to trigger jenkins to decide if the build could go to production or not. So the pipline looks as below.
After deploying to uat, I can send the webhook to trigger e2e test on gitlab. And gitlab can send the webhook to jenkins, however, as the pipeline job e2e received the webhook and build success. It does not change the pipeline job status, and hence, we can not proceed to deploy to production unless we manually click the trigger on e2e job. And I already try to use currentBuild.currentResult and currentBuild.result which doesn't seem to work.
What should I do to solve this?
Update
I am not using the new pipeline way with groovy, instead, I'm doing it the old way like this. But I supposed this is not related to the issue I'm trying to solve here.
Let me provide more details about the problem. After e2e test is done, I send below request from gitlab to jenkins
curl "SERVER_URL/job/MY_XXX_PROJECT/view/Pipeline/job/X.X%20E2E%20my_e2e_test_result/buildWithParameters?token=xxxxxxxxxxxxxx&PARENT_BUILD_NUMBER=123¤tBuild.currentResult=SUCCESS¤tBuild.Result=SUCCESS"
And here is the pipeline, and if I check the job in SERVER_URL/job/MY_XXX_PROJECT/view/Pipeline/job/X.X%20E2E%20my_e2e_test_result/. The job is actually built SUCCESS.
However, the status shows like nothing happened, and we cannot proceed to deploy production without manually clicking the trigger button on the same job(e2e) again.
If you're just looking for a way for the job to update its own state, use the unstable and the error steps.
pipeline {
agent any
stages {
stage('Hello') {
steps {
echo 'This job is now a success. All jobs are a success until marked otherwise'.
//success 'This will fail. There's no such thing as a success step'
unstable 'This stage is now unstable'
error 'This job is now failed'
unstable 'This job is still failed. Build results can only get worse. They can never be improved.'
}
}
}
}
Notice that there's no such thing as a success step. Probably because all jobs are a success until told otherwise. And the overall result of a job cannot be improved. See the documentation for catchError here.
Note that the build result can only get worse, so you cannot change the result to SUCCESS if the current result is UNSTABLE or worse.
I have created a jenkins pipeline job called "pipelinejob" with the below script:
pipeline {
agent any
stages {
stage ('Setup'){
steps{
//echo "${BRANCH_NAME}"
echo "${env.BRANCH_NAME}"
//echo "${GIT_BRANCH}"
echo "${env.GIT_BRANCH}"
}
}
}
}
Under General, I have selected "GitHub project" and inserted my company's github in the form:
https://github.mycompany.com/MYPROJECTNAME/MY_REPOSITORY_NAME/
Under Build Triggers, i have checked "GitHub hook trigger for GITScm polling
I have created a simple job called "simplejob" with same configuration as 1) and 2)
In my company's Github, i have created a webhook like "jenkins_url/jenkins/github-webhook/"
I commit a change in "mybranch" in "MY_REPOSITORY_NAME"
My simple job "simplejob" is triggered and built successfully
My pipeline job "pipelinejob" is not triggered
In Jenkins log i see the below:
Sep 12, 2019 2:42:45 PM INFO org.jenkinsci.plugins.github.webhook.subscriber.DefaultPushGHEventSubscriber$1 run
Poked simplejob
Nothing regarding my "pipelinejob".
Could you please point me to the right directions as to what to check next?
P.S. I have manually executed my "pipelinejob" successfully
I wasted two days of work on this, as none of the previous solutions worked for me. :-(
Eventually I found the solution on another forum:
The problem is that if you use a Jenkinsfile that is stored in GitHub, along with your project sources, then this trigger must be configured in the Jenkinsfile itself, not in the Jenkins or project configuration.
So add a triggers {} block like this to your Jenkinsfile:
pipeline {
agent any
triggers {
githubPush()
}
stages {
...
}
}
Then...
Push your Jenkinsfile into GitHub
Run one build manually, to let Jenkins know about your will to use this trigger.
You'll notice that the "GitHub hook trigger for GITScm polling" checkbox will be checked at last!
Restart Jenkins.
The next push should trigger an automated build at last!
On the left side-pane of your pipeline job, click GitHub Hook log. If it says 'Polling has not run yet', you will need to manually trigger the pipeline job once before Jenkins registers it to poke on receiving hooks.
Henceforth, the job should automatically trigger on GitHub push events.
I found an answer to this question with scripted pipeline file. We need to declare the Github push event trigger in Jenkins file as follows.
properties([pipelineTriggers([githubPush()])])
node {
git url: 'https://github.com/sebin-vincent/Treasure_Hunt.git',branch: 'master'
stage ('Compile Stage') {
echo "compiling"
echo "compilation completed"
}
stage ('Testing Stage') {
echo "testing completed"
echo "testing completed"
}
stage("Deploy") {
echo "deployment completed"
}
}
}
The declaration should be in the first line.
git url: The URL on which pipeline should be triggered.
branch: The branch on which pipeline should be triggered. When you specify the branch as master and make changes to other branches like develop or QA, that won't trigger the pipeline.
Hope this could help someone who comes here for an answer for the same problem with Jenkins scripted pipeline :-(.
The thing is whenever you create a pipeline job for git push which is to be triggered by github-webhook, first you need to build the pipeline job manually for one time. If it builds successfully, then Jenkins registers it to poke on receiving hooks. And from the next git push, your pipeline job will trigger automatically.
Note: Also make sure that the pipeline job built manually for the first time should be built successfully, otherwise Jenkins will not poke it. If it fails to build, you can never trigger the job again.
I have an upstream pipeline which is calling another downstream pipeline
build job: "/org/projectA/master",
parameters: [[$class: 'StringParameterValue', name: 'variable', value: 'value']],
wait: true
In my downstream pipeline, there is a step to ask for approve
input "Deploy to prod?"
Currently the job is paused in the downstream pipeline waiting for approve, but in my main job (upstream pipeline), it is just waiting for sub pipeline to finish, doesn't show any message for approver. So is it possible to display the interactive input in my main pipeline? then the approver doesn't need to click to the sub pipeline to check the status.
BTW, I cannot move the input to main pipeline, cause there are other steps after it in the sub pipeline.
Thanks in advance for any suggestion
I really wouldn't recommend it, but there's a way via Jenkins Remote API -
Jenkins input pipeline step filled via POST with CSRF - howto?
curl -X POST -H "Jenkins-Crumb:${JENKINS_CRUMB}" -d json='{"parameter": {"name": "${PARAMETER_NAME}", "value": "${PARAMETER_VALUE}"}}' -d proceed='${SUBMIT_CAPTION}' 'http://j${JENKINS_URL}/job/${JOB_NAME}/${BUILD_ID}/input/${INPUT_ID}/submit'
The question would be how would you run this? A new input in the upstream job? Run when?
It might be more useful to divide the downstream job into two and run the actual deploy only when user accepts the input in the upstream job.
Executing upstream job called "A". On success of A executing test cases which is downstream project "B". But while sending mail from B we have to incorporate upstream project details (upstream project name, build no) in mail. So we can easily map / corelate the test run with respective upstream job.
In downstream project dashboard below details are displaying.
Started by upstream project Dev_RM_3.0_CI_Test build number 10
originally caused by:
I checked in https://wiki.jenkins-ci.org/display/JENKINS/Building+a+software+project. but couldnt find anything to inherit in downstream.
Created sample job with below details to display the current job details.
echo $BUILD_NUMBER
echo $JOB_NAME
echo $BUILD_ID
But the output is
Building on master in workspace /var/lib/jenkins/workspace/env
[env] $ /bin/sh -xe /tmp/hudson970280339057643719.sh
+ echo 1
1
+ echo env
env
+ echo 1
1
Finished: SUCCESS
Any help to inherit upstream details in downstream job?
How to get current job details?
The message that you refer to your question "Started by upstream project "Chained/1-First" build number 34" for example, is available in the jenkins Cause.
Jenkins keeps the upstream build info in it's cause object. If your are using build DSL or Pipelines you may get it in groovy. Alternatively you can curl the job url and use jq to get the Cause
For example curl http://localhost:8080/job/Chained/job/2-Second/17/api/json
"_class": "org.jenkinsci.plugins.workflow.job.WorkflowRun",
"actions": [{
"_class": "hudson.model.CauseAction",
"causes": [{
"_class": "hudson.model.Cause$UpstreamCause",
"shortDescription": "Started by upstream project \"Chained/1-First\" build number 34",
"upstreamBuild": 34,
"upstreamProject": "Chained/1-First",
"upstreamUrl": "job/Chained/job/1-First/"
}]
}
Or from the pipeline for example:
node() {
stage('downstream') {
def upstream = currentBuild.rawBuild.getCause(hudson.model.Cause$UpstreamCause)
echo upstream?.shortDescription
}
}
You can get a bunch of information out of Cause, pending all the script approvals or a global shared step. You will get a null if a different cause triggers this build, eg commit, or user.
You can pass in the upstream variables via build parameters to the downstream job and then you can access them (in the downstream job) using things such as ${MyParameter1} and ${MyParameter2}.
You would need to:
Add build parameters to the downstream job. For example, a string parameter named "ParentJobName".
Add a post build "Trigger downstream parameterized builds on other projects" to the upstream job.
Add something like "Current Build parameters" or "Predefined parameters" to the #2 and pass in whatever you need. For example:
ParentJobName=${JOB_NAME}
Access the parameters as you would other build variables. e.g. ${ParentJobName}
You should be able to pass in the basic stuff that way. Anything more complicated than that and you will probably be better off using a plugin like Copy Artifacts Plugin to copy files or using the Jenkins API in a system groovy step to get/modify the upstream build, etc.
You can simply use params.variableName in your downstream job to retrieve the parameters passed from your upstream parameter job. Your downstream job need not necessarily be a parameterized job.
Extending #razboy answer:
this is good way if Cause cannot be whitelisted in sandbox. I forgot about Jenkins API and used current build console to look for string about trigger cause. You can try to fetch data from API as #razboy or get current console and grep it if you need simple stuff. Jenkins API is more flexible for more complex logic. To get API help,append /api to your build url: <jenkins_url>/job/<buildUrl>/<buildNumber>/api
def buildUrl = env.BUILD_URL
sh "wget $buildUrl -O currentConsole.txt"
statusCode = sh returnStatus: true,script: 'cat currentConsole.txt | grep -q "Started by upstream project"'
boolean startedByUpstream= statusCode==0
MeowRude's answer helped me. To repcap it, in upstream job:
build job: 'mail-test', parameters: [[$class: 'StringParameterValue', name: 'VERSION_NUMBER', value: '1.0.0.0']]
And in downstream job:
echo "${params.VERSION_NUMBER}"
You may have to have certain plugins installed, but
def causes = currentBuild.getBuildCauses()
will return an ArrayList of objects that will most likely provide the necessary details, for example upstreamProject for the full project name and upstreamBuild for the build number. Then you can correlate results between up- and downstream builds easily.
Source: link to pipeline-examples in razboy's comment above
I'm new to Jenkins and I've been tasked with a simple task of passing the output from one pipeline to the other.
Lets say that the first pipeline has a script that says echo HelloWorld, how would i pass this output to another pipeline so it displays the same thing.
I've looked at parameterized triggers and couple of other answers but I was hoping if someone could layout the step by step procedure to me.
If you want to implement it purely with Jenkins pipeline code - what I do is have an orchestrator pipeline job that builds all the pipeline jobs in my process, waits for them to complete then gets the build number:
Orchestrator job
def result = build job: 'jobA'
def buildNumber = result.getNumber()
echo "jobA build number : ${buildNumber}"
In each job like say 'jobA' I arrange to write the output to a known file (a properties file for example) which is then archived:
jobA
writeFile encoding: 'utf-8', file: 'results.properties', text: 'a=123\r\nb=foo'
archiveArtifacts 'results.properties'
Then after the build of each job like jobA, use the build number and use the Copy Artifacts plugin to get the file back into your orchestrator job and process it however you want:
Orchestrator job
step([$class : 'CopyArtifact',
filter : 'results.properties',
flatten : true,
projectName: 'jobA',
selector : [$class : 'SpecificBuildSelector',
buildNumber: buildNumber.toString()]])
You will find these plugins useful to look at:
Copy Artifact Plugin
Pipeline Utility Steps Plugin
If you are chaining jobs instead of using an orchestrator - say jobA builds jobB builds jobC etc - then you can use a similar method. CopyArtifacts can copy from the upstream job or you can pass parameters with the build number and name of the upstream job. I chose to use an orchestrator job after changing from chained jobs because I need some jobs to be built in parallel.