Finally block is cancelled if task block timeouts and global pipeline timeout is not exceeded

[wilstdu]

Expected Behavior

Finally runs when tasks timeout. I expect to see a running finally block if tasks timeout and the global pipeline timeout is not exceeded.

Actual Behavior

Finally doesn't run when tasks timeout.

Steps to Reproduce the Problem

1. Create a simple pipeline that has a longer running tasks part and a finally block with any task
2. Configure PipelineRun timeouts, so that it would have timeout for tasks and finally parts

Some test cases from me:

timeouts:
  tasks: 0h1m0s
  finally: 0h0m20s

# PipelineRun "code-analyzer-106" failed due to tasks failed to finish within "1m0s"

  

timeouts:
  pipeline: 0h2m0s
  finally: 0h0m20s

# PipelineRun "code-analyzer-105" failed due to tasks failed to finish within "1m40s"

timeouts:
  pipeline: 0h2m0s
  tasks: 0h1m0s
  finally: 0h0m20s
 # PipelineRun "code-analyzer-104" failed due to tasks failed to finish within "1m0s"

Additional Info

• Kubernetes version: 1.31

• Tekton Pipeline version: v1.2.0

[waveywaves]

/assign