Post History

How to manage CPU capabilities of Jenkins nodes?

We use Jenkins Pipeline to build and test some C++ software. The pipeline script runs the tests on 10 different nodes in a parallel way in order to save time. All of these nodes are real (not virtual) computers on the network.

Some of the computers have old CPUs, which don't support modern instruction sets (e.g. AVX2). I want to use the correct build script for each node - if it supports AVX2, use the corresponding script, and if not, use a fall-back script.

I wonder how I should manage that in the Jenkins controller and in my Jenkins Pipeline script. I am not allowed to install programs like [Coreinfo](https://docs.microsoft.com/en-us/sysinternals/downloads/coreinfo) on the nodes, so the Jenkins Pipeline script cannot determine directly which build script to run. Instead of that, I imagine the Jenkins controller could support some "properties" for each node, which I could set manually, using trial and error (i.e. try to run the AVX2 script; if it fails, mark the node as "old"). I tried but couldn't find such a feature. Does Jenkins have it?

If not, which other feature could I use to record such information for each node? How can the Pipeline script then get this information?

Below is a sketch of the Pipeline script we use. It uses "label" to force the system to run it on a specific node.

```
pipeline {
    agent {label 'specific-computer-name'}
    stages {
        stage('build') {
            steps {
                bat 'whatever'
            }
        }
        stage('test') {
            steps {
                bat 'whatever'
            }
        }
    }
}
```