Understanding your Jenkins CI/CD Pipeline
Structure
The Jenkins DevOps pipeline is defined in a ‘Jenkinsfile’ (with the default filename Jenkinsfile
) and has an overall structure summarised below. Note that paired curly braces - { }
- denote the start and end of a scoped section. Comments can be created in a block (/* comment like this */
) or inline (// comment like this
).
pipeline { # Outermost scope
agent { ... } # What are the default agent(s) upon which Stages execute?
parameters { ... } # What values are required if the pipeline is invoked interactively?
environment { ... } # What (global) variables are needed?
stages { # Operational container
stage { ... } # First Stage (sequence of executable steps)
agent { ... } # Which agent is used to execute this Stage?
environment { ... } # What variables are needed? (local to this stage)
steps {
{ step } # Executable step
...
} # End of first Stage
stage { ... } # Second Stage
...
} # End of Stage definitions
} # End of Pipeline
Note that Jenkins offers some flexibility in the structure of the pipeline, and in some cases allows repetition at different scopes. For example, a single Step can optionally contain another Stages {...}
container which defines more sub-Stages). To understand these sections in more detail refer to the Jenkins documentation.
Input Values
Parameters section
The parameters
section enumerates the parameters to the pipeline with their default values. Note that parameters values are immutable, and once a pipeline is invoked with parameter values those values cannot be changed.
Note that the sample Jenkins pipelines supplied with MettleCI do not make use of parameters as all pipeline values are supplied as Node Properties.
Environment Section
The environment
section allows you to derive and set further useful variables. The template pipeline composes the suffixed DataStage project and copies one of the parameters to an all uppercased variable for convenience later.
environment {
DATASTAGE_PROJECT = "${params.projectName}_${params.environmentId}"
ENVIRONMENT_ID = "${params.environmentId}"
}
These will be visible in the execution environment as environment variables, and thus will be available to batch files, external commands, and the like without needing to explicitly pass them.
Environment variables can be further manipulated in the same or subsequent environment {}
section(s)
Stages Section
The Stages section (see terminology) is a container for an arbitrary number of individual Stage containers, which
Contains Steps - the executable actions
which are normally run in sequence (but can be configure to run in parallel).
Stage section notes
Each stage section begins with a label, which is a quoted arbitrary text string displayed on the pipeline diagram in Blue Ocean when viewing the pipeline,
Within the stage are two major subsections, steps{} and post{}, which contain the actions to be performed during normal execution, and the actions to be performed at the end of the stage
Use of the parallel { } construct allows nesting of stages within stages.
Deploy stage
This is the first stage in the pipeline and it is tasked with deploying changed project assets to the CI (continuous integration) project. WithCredentials
provides appropriate credentials (using the credentials plugin) to the rest of the steps. change from mci-user if you are using a different user. Each bat step invokes the MettleCI CLI to perform one of the deployment tasks. Refer to the MettleCI - Auto-Generated CLI Syntax Reference for details of what each step is doing. The label, chosen to be descriptive of the step task, becomes the step text in the Blue Ocean display of the pipeline. If a step aborts that ends the steps part of the stage. After the steps conclude, by error or by reaching the end, the post section is evaluated. Although there are other values such as changed, fixed, etc, using always ensures the section is always executed to process test results (here, the result of the compilations) and then cleanup.
stages {
stage("Deploy") {
steps {
withCredentials([usernamePassword(credentialsId: 'mci-user', passwordVariable: 'datastagePassword', usernameVariable: 'datastageUsername')]) {
bat label: 'Create DataStage Project', script: "${env.METTLE_SHELL} datastage create-project -domain ${params.domainName} -server ${params.serverName} -project ${env.DATASTAGE_PROJECT} -username ${datastageUsername} -password ${datastagePassword}"
bat label: 'Substitute parameters in DataStage config', script: "${env.METTLE_SHELL} properties config -baseDir datastage -filePattern \"*.sh\" -filePattern \"DSParams\" -filePattern \"Parameter Sets/*/*\" -filePattern \"*.apt\" -properties var.${params.environmentId} -outDir config"
bat label: 'Transfer DataStage config and filesystem assets', script: "${env.METTLE_SHELL} remote upload -host ${params.serverName} -username ${datastageUsername} -password ${datastagePassword} -transferPattern \"filesystem/**/*,config/*\" -destination \"${env.DATASTAGE_PROJECT}\""
bat label: 'Deploy DataStage config and file system assets', script: "${env.METTLE_SHELL} remote execute -host ${params.serverName} -username ${datastageUsername} -password ${datastagePassword} -script \"config\\deploy.sh\""
bat label: 'Deploy DataStage project', script: "${env.METTLE_SHELL} datastage deploy -domain ${params.domainName} -server ${params.serverName} -project ${env.DATASTAGE_PROJECT} -username ${datastageUsername} -password ${datastagePassword} -assets datastage -parameter-sets \"config\\Parameter Sets\" -threads 8 -project-cache \"C:\\dm\\mci\\cache\\${params.serverName}\\${env.DATASTAGE_PROJECT}\""
} // end of withCredentials section
} // end of steps in this stage
post {
always {
junit testResults: 'log/**/mettleci_compilation.xml', allowEmptyResults: true
withCredentials([usernamePassword(credentialsId: 'mci-user', passwordVariable: 'datastagePassword', usernameVariable: 'datastageUsername')]) {
bat label: 'Cleanup temporary files', script: "${env.METTLE_SHELL} remote execute -host ${params.serverName} -username ${datastageUsername} -password ${datastagePassword} -script \"config\\cleanup.sh\""
} // end of things to "always" perform
} // end of post steps section
} // end of deploy stage
Test stage and parallel Static Analysis and Unit Tests substages
After a successful deployment the pipeline will perform some tests, compliance and the unit tests. These can be performed in parallel as there is no interdependence, so we use the parallel{}
construct to nest the stage invocations. As before, refer to the MettleCI - Auto-Generated CLI Syntax Reference for details of what each step is doing.
Compliance testing is relatively simple, requiring only a single step to run the rules with no cleanup, and subsequent collection of the results. Unit tests are a bit more involved with upload of the specs, execution of the jobs in test mode, and downloading the reports followed again by collection of results
Some notes:
The template assumes the compliance rules are in the same repo. If one wishes to source the rules from a different repo and further, to run separate warn/fail sets of tests, presenting the result in one report, the static analysis stage would look like this instead: (refer to Jenkins documentation to understand the parameters in the checkout plugin invocation)
Promotion stage with three parallel deployment substages
The template pipeline has three promotion deployments, each of which is essentially the same as the initial deploy but directed at a different project. Our template assumes that all three deployments are to the same DataStage server, but the parameter blocks are at the individual promotion level so you can change these as necessary. You can reduce or increase (by deleting or copying code blocks) the number as necessary. These can be performed in parallel as there is no interdependence, so we again use the parallel{}
construct to nest the stage invocations. As before, refer to the MettleCI - Auto-Generated CLI Syntax Reference for details of what each step is doing. The “innards” of each promotion step, save the first are elided in the following code snippet to focus on the structure.
The input{} construct is used to gather parameters and get approval from the user for the non automatic steps. In the template, all three are automatic.
As in the test stage, we take advantage of the ability to nest stages using the parallel{} construct. The second and third nested stage are similar
If we had wished to have the first (“testing”) stage run without prompting the user, we would change it to look like as follows (steps omitted for brevity)
Many other variations are possible. Consult with your Jenkins resources for more nuances.
© 2015-2024 Data Migrators Pty Ltd.