I'm using Jenkins and in particular pipelines. What's the difference between pipelines and Hudson jobs with multiple build steps?
When Kohsuke forked Jenkins from Hudson the entire developer community went with him. Oracle's grab for Hudson and attempt to make it proprietary killed Hudson overnight. This is well documented elsewhere. No one in their right mind should be using Hudson over Jenkins.
Hudson jobs are just freestyle jobs with a lot less functionality because Jenkins has, since forking, greatly increased the number of plugins as it had the developer community. Oracle had one guy sitting in a corner hacking on Hudson.
For one thing, Pipeline jobs allow you to version your job definitions in source control. Old school Freestyle jobs do not. To list all the other ways Jenkins is better than Hudson would require a million monkeys writing the works of Shakespeare!
Related
We are starting to develop CI workflow for our systems in my company.
Currently we just making few basic tasks like build, tests, and upload to Nexus.
The tech stack is a Java project which build in Gradle and Jenkins makes our build.
Currently i'm working with some basic Groovy script to make what we need, but each time i'm copy and paste my updated code to Jenkins and running the job from Jenkins UI to see the results, and to me it seems like not a very good approach for developing such automation code.
My question is, what is the best practice to build and run Jenkins jobs?
Is it possible to run it straight from Intellij ?
Do we need to create a Jenkins project which should be saved as a repository and then deploy it to Jenkins machine?
Do we need to use some Intellij plugins in order to work with Jenkins?
More best practices are welcome :)
Jenkins has an API - so you can do whatever you want!
But in general, for small to medium teams it's better to use Jenkinsfile and let Jenkins pull code changes (or pull-requests) from SCM and trigger builds. You can also configure hooks to trigger builds if your SCM supports this (Github & bitbucket supports this).
If you are eventually pushing your artifacts to a docker image, I would highly recommend docker multi-stage builds.
If you are completely new to CI/CD stuff - Atlassian has a lot of good resources https://www.atlassian.com/continuous-delivery/principles/continuous-integration-vs-delivery-vs-deployment
I have a 4-5 java projects build procedure to be configured as CI using Jenkins. Whether it will be time saving to build some/all projects on different machines(connected as Jenkins slaves)?
Are there some any other benefits of Jenkins Master-Slave configuration?
Offloading work to build agents is a very good idea as it keeps load away from the master. This allows you to build more projects in parallel (esp. with dynamic agents launched in some cloud environment).
Further, it makes the systems easier to maintain, as e.g. the Java version/setup of your build agents required to build and test your application does not interfere with the Jenkins master machine.
I am trying to understand how to best deploy an instance of Jenkins, complete with plugins, users and jobs using Chef. I am currently using the Chef Jenkins Supermarket cookbook.
I am attempting to achieve automated deployment of our Pipelines as part of the project. From what I have gathered, the best way to go about this is to have Chef configure a seed job in Jenkins initial setup and configuration.
The seed job should specify, among other things, the git repository from which to find and use a Jenkinsfile for a given job. I've found this resource by Daniel Spilker to be helpful in explaining seed jobs.
So the seed Jenkins job would be run, which would then generate the Jenkins job we have just scripted with it (in this case the seed job would be to pull the Jenkinsfile from source control and configure a new Jenkins job (our pipeline), with the details of the Jenkinsfile).
Am I understanding this correctly as the proper way to not only automate Jenkins job configuration, but also as the proper way to always have an up to date job configuration for any given job in the event the job configuration were to change?
If we used a seed job to setup our pipeline, what are some possible solutions to having the initial seed job run automatically once Jenkins is fully configured by Chef?
As for job configuration changes that may occur over time, would we need to setup the seed job to poll source control for any changes in the Jenkinsfile periodically in the event the Jenkinsfile has been modified? (It may be helpful to note that we are currently using BitBucket for source control).
Just getting started with pipeline as code. Thanks to everybody in advance for their patience and guidance.
I've mentioned this a bit in your other questions, but the least painful approach is to treat Jenkins as a database, not a web service. Have Chef do the basic install, but then configure the initial bits by hand. For DR, rely on your backups rather than Chef.
Currently, I use the Build Pipeline Plugin to orchestrate the delivery of my code through the different environments:
Build the code and execute unit tests
Manually deploy to the development environment
Automatically execute tests on the development environment
Manually release the software and put the version number to the released version.
Manually deploy to the integration test environment by downloading the artefact from a repository, based on the version put by the release build.
Manually deploy to ...
With Jenkins 2.0 comes the Pipeline plugin. But how do these two plugins relate to each other?
Should I migrate to the latest plugin? The things I seem to miss from the Jenkins 2 Pipeline plugin:
Manually trigger a stage. I can wait for an input, but it does not seem to be so elegant
Restart a stage to retrigger a deployment. This does not seem possible.
Visibility into the parameters that were used to trigger a stage, e.g. the version number of the software that was deployed.
Am I missing the point here? Should the two of them be combined? Or how are you approaching a pipeline like this?
With the current state of Jenkins 2 pipelines you are correct to state all the 'missing features' you listed.
One of the advantages of the Jenkins 2 pipeline plugin is that rather than chaining together a series of jobs as with the Build Pipeline Plugin, your entire pipeline is 1 'job', which makes user administration much easier IMO.
The other advantage of Jenkins 2 pipelines would be 'configuration as code', so you can track changes to your pipeline as you would any other file in version control.
Jenkins 2 pipelines are very much the new 'hotness', and there's many plugins implementing compatability day after day.
Once the new UI becomes production ready, I'd imagine that the old build pipeline plugin will begin to be deprecated.
Also you should be aware that the Build Pipeline plugin is not maintained by the Jenkins or CloudBees teams as far as I know, whereas Jenkins 2 pipelines are.
Would I recommend migrating now? No, I personally still don't consider the Jenkins 2 pipelines mature enough for deployment to production in an organisation. I'd stay to stick with what you know for now while you wait for the Jenkins 2 Pipeline ecosystem to mature.
My reasoning I gave in a blog post a few weeks ago (read more here if you want, but I've extracted out the 'weaknesses' here for you):
There are still plenty of plugins that I and many others will consider 'core to their CI pipeline' missing full or partial support for pipelines.
The lack of 'per-project-configuration' in pipelines for many plugins. e.g. Slack - the current implementation 'assumes' that all Jenkins 2 Pipeline projects should be communicated to the same Slack channel/team - whereas you may want to have multiple Slack teams configured. There are multiple other plugins like this.
At present the documentation of Jenkins 2 Pipelines is very limited, though this is improving.
I have a Jenkins pipeline which is responsible for about 5 stages (build and several different tests). I'm migrating from Jenkins 1.XX (with no pipelines) to Jenkins 2 and I'd like to replicate my process as closely as possible. The pipeline job I have set up on J2 handles everything exactly the same except it's using a JenkinsFile. The only issue is that the pipeline plugin does not appear to support Promotion of particular builds like you can do with Freestyle Jobs. Has anyone found a way around this?
Seems like Promoted Builds Plugin doesn't support pipelines yet. If you check the issue there is also some good pros and cons about supporting it for pipelines.