NVIDIA / gpu-operator / 38674752672

45%

main: 45%

Build Type

Pull #3081

github

Committed by tariq1890

Commit Message

fix[device-plugin] add cdi validation guardrail

This commit adds additional validation to the devicePlugin operand
transformation logic. When CDI is enabled, it is required that the
DEVICE_LIST_STRATEGY includes one of "cdi-annotations" and "cdi-cri".
While the current transformation code adds these values to DEVICE_LIST_STRATEGY
when .spec.cdi.enabled is true, the same resolved value gets overridden
when the gpu-operator user sets the DEVICE_LIST_STRATEGY env var explicitly
in the devicePlugin section of the ClusterPolicy custom resource. This
leads to an invalid devicePlugin config whose failure only manifests when
pod workloads fail during the "containerCreate" stage. This provides a poor
UX as users waste hours debugging the containerCreate failures instead of
identifying the invalid devicePlugin config. This change changes that by
failing at the ClusterPolicy reconcile stage.

Signed-off-by: Tariq Ibrahim <tibrahim@nvidia.com>

Pull Request Pull Request #3081: fix[device-plugin] add cdi validation guardrail

Coverage Stats

8 of 11 new or added lines in 1 file covered. (72.73%)

439 existing lines in 1 file now uncovered.

5867 of 13036 relevant lines covered (45.01%)

0.45 hits per line