Description of problem:

We are observing intermittent failures of Tekton PipelineRuns where the PipelineRun transitions to Failed with the message:

Tasks Completed: 4 (Failed: 1, Cancelled 0), Skipped: 14

even though all completed TaskRuns show Succeeded, and no task is explicitly cancelled.

This issue has been reproduced multiple times in the openstack-tenant namespace under high concurrency and appears to be a controller-side race condition rather than a user configuration or task failure.

Prerequisites (if any, like setup, operators/versions):

Actual results:

• PipelineRun transitions to Failed while a TaskRun is still Running

• Remaining tasks are skipped automatically

• Failure reason does not correspond to any real TaskRun failure

Expected results:

• PipelineRun should remain in Running state while any TaskRun is still running

• Downstream tasks should not be skipped unless:

• • A TaskRun has explicitly failed

• • PipelineRun is explicitly cancelled or timed out

Reproducibility (Always/Intermittent/Only Once):

yes

Suspected Root Cause

A race condition in the Tekton Pipelines controller during reconciliation under high concurrency, where a transient or stale TaskRun state is incorrectly interpreted as a failure, triggering PipelineRun stopping logic prematurely.

Acceptance criteria: 

Definition of Done:

Build Details:

Additional info (Such as Logs, Screenshots, etc):

Slack Thread - https://redhat-internal.slack.com/archives/C04PZ7H0VA8/p1768478363451539

 *