Description

Upstream Issue: https://github.com/tektoncd/pipeline/issues/8495

PipelineRuns are being reconciled thousands of times repeatedly while in progress, even when there are no changes. This causes excessive load on the controller and impacts performance.

Expected Behavior

Reconciliation should only happen when:

• The PipelineRun changes
• Child TaskRuns change
• Other relevant resources change

Actual Behavior

The controller reconciles the same PipelineRun thousands of times continuously while it's in progress. Eventually the PipelineRun completes successfully, but the excessive reconciliation causes:

• High CPU usage on the controller
• Increased latency for other operations
• Potential scalability issues

Impact

• Performance degradation on clusters with many PipelineRuns
• Controller resource exhaustion
• Affects scalability and production stability
• May impact SLAs for pipeline execution times

Evidence

Controller logs show repeated reconciliation of the same PipelineRun without changes. See attached logs in the upstream issue.

Environment

• Tekton Pipelines: v0.66.0
• Kubernetes: v1.30.6

Notes

Investigation needed to identify root cause and reproduce reliably.