Epic Goal

Provide experiences that let developers control how builds are scheduled on clusters, including:

• Specifying nodes with node selectors. Spun out to BUILD-1150
• Tolerating node taints.
• Instructing Kubernetes to use a custom scheduler when evaluating node placement.

Why is this important?

• Builds may need to tolerate certain node taints if cluster admins want to steer builds to specific nodes.
• Cluster admins may want builds to use a custom scheduler which is optimized for Tekton workloads (or similar for batch, AI/ML, etc.).

Scenarios

1. Clusters running custom schedulers for batch/AI workloads
2. Clusters whose admins use node taints to segregate workloads

Acceptance Criteria (Mandatory)

• Builds can be instructed to tolerate specific node taints
• Builds can be scheduled with a custom scheduler
• Tests verify behavior upstream and downstream
• Hardening guideline published to to mitigate potential system availability issues that could be exposed by these features.

Dependencies (internal and external)

1. Tekton support for Pod Template tuning.

Previous Work (Optional):

1. BUILD-972 - enhancement proposal for build scheduler options (see also SHIP-0039)

Open questions::

1. Do we have a preferred admission controller for OpenShift? Ex: OPA Gatekeeper, Kyverno, etc?

Done Checklist

• Acceptance criteria are met
• Non-functional properties of the Feature have been validated (such as performance, resource, UX, security or privacy aspects)
• User Journey automation is delivered
• Support and SRE teams are provided with enough skills to support the feature in production environment