#7153 Use retryable failure for failed Ray job deployments Pull request opened by pingsutw Tracking issue Why are the changes needed? When a Ray job deployment fails, the plugin should return a retryable failure (system error) instead of a permanent failure. This allows Flyte to retry the task, which is the desired behavior since deployment failures are often transient (e.g., cluster scheduling issues, temporary resource constraints). What changes were proposed in this pull request? • Changed

JobDeploymentStatusFailed

handler from

PhaseInfoFailure

to

PhaseInfoSystemRetryableFailureWithCleanup

so failed Ray jobs are retried • Added

TestBuildResourceRayMissingPrimaryContainer

- validates error when primary container is missing from pod spec • Added

TestGetTaskPhaseFailedRetryable

- validates retryable failure phase with reason/message propagation • Updated

TestGetTaskPhase

to expect

PhaseRetryableFailure

for failed deployments How was this patch tested? Unit tests added and all existing tests pass:

go test ./go/tasks/plugins/k8s/ray/ -count=1

Labels • changed: Ray job failure handling changed to retryable • added: New unit tests Check all the applicable boxes • I updated the documentation accordingly. • All new and existing tests passed. • All commits are signed-off. flyteorg/flyte