配置Jenkins在构建时根据构建参数以编程方式确定从属？

Question

This is perhaps a slightly unusual Jenkins query, but we've got a project that spans many projects. All of them are Linux based, but they span multiple architectures (MIPS, SPARC, ARMv6, ARMv7).

For a specific component, let's call it 'video-encoder', we'll therefore have 4 projects: mips-video-encoder, sparc-video-encoder, etc.

Each project is built on 4 separate slaves with a label that correlates to their architecture, ie the MIPS slave has the labels 'mips' 'linux'.

My objectives are to:

1. Consolidate all of our separate jobs. This should make it easier for us to modify job properties as well as easier to add more jobs without the duplicitous effort of adding so many architecture specific jobs.
2. To allow us to build only one architecture at a time if we so wish. If the MIPS job fails, we'd like to build just for MIPS and not for others.

I have looked at the 'Multi-configuration' type job -- at the moment we are just using Single confguration jobs which are simple. I am not sure if the Multi-configuration type allows for us to build only individual architectures at once. I had a play with the configuration matrix, but wasn't sure if this could be changed / adapted to just build a for single platform. It looks like I may be able to use a Groovy statement to do this? Something like:

(label=="mips".implies("slave"=="mips")

Maybe that could be simplified to something like slave == label where label is the former name of the job when it was in its single-configuration state and is now a build parameter?

I am thinking that we don't need a Multi-config job for this, if we can programatically choose the slave for this.

I would greatly appreciate some advice on how we can consolidate the number of jobs we have and programatically change the target slave based on the architecture of the project which is a build parameter.

Many thanks in advance,

Answer 1

You can make a wrapper job with a system groovy script. You need the groovy plugin for this. let call the wrapper job - video-encoder-wrapper , here are the bullets how to configure it:

• Define the parameter ARCH

• Assign the label to the video-encoder job based on the ARCH parameter by the step Execute system Groovy script

   import hudson.model.* encoder=Hudson.instance.getItem('video-encoder') def arch =build.buildVariableResolver.resolve("ARCH") label= Hudson.instance.getLabel(arch) encoder.setAssignedLabel(label) 

• Invoke non blocking downstream project video-encoder , don't forget to pass the ARCH parameter

• Check the option Set Build Name in the video-encoder job's configuration and set it to the something as ${ENV,var="ARCH"} - #${BUILD_NUMBER} . It will allow you to track easily the build history.
• Disable the concurrent builds of video-encoder-wrapper job. It will prevent the assigning of 2 different labels in the same time to the video-encoder job

Hope it helps