How to disable the default Node Affinity rule added to Kubernetes agents of CloudBees CI ?

Article ID:360045777812
1 minute readKnowledge base

Issue

  • I want to disable the node affinity rule that is automatically added to CloudBees CI (CloudBees Core) kubernetes agents:

kind: "Pod"
spec:
  affinity:
    podAntiAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
        - weight: 1
          podAffinityTerm:
            topologyKey: kubernetes.io/hostname
            labelSelector:
              matchExpressions:
                - key: com.cloudbees.cje.type
                  operator: In
                  values:
                    - master
                    - cjoc

Environment

Explanation

Since version 1.11.17 of the CloudBees Kube Agent Management plugin, a node affinity rule is automatically added to the agents spun up. This rule gives a scheduling preference to nodes where there are no controllers or Operations Center running.

Resolution

The default affinity rule can be disabled via a System Property. Depending on the version of the CloudBees Kube Agent Management plugin, the system property differs. See How to add Java arguments to Jenkins on CloudBees CI Modern ? to understand how to do this.

Version 1.1.32 and later

Add the system property com.cloudbees.jenkins.plugins.kube.Config.enableAffinity=false and restart the controller from Operations Center.

The rule can also be disabled live by executing the Groovy script com.cloudbees.jenkins.plugins.kube.Config.ENABLE_AFFINITY=false under Manage Jenkins  Script Console. This does not survive a restart however.

Version prior to 1.1.32

Add the system property com.cloudbees.jenkins.plugins.kube.AffinityPodTemplateFilter.disabled=true and restart the controller from Operations Center.

This article is part of our Knowledge Base and is provided for guidance-based purposes only. The solutions or workarounds described here are not officially supported by CloudBees and may not be applicable in all environments. Use at your own discretion, and test changes in a safe environment before applying them to production systems.