NVIDIA / gpu-operator / 39687013990

45%

main: 37%

push

github

Address review: count nodes early, log on panic, emit ClusterPolicy start

The GPU node gauge was set after the prerequisite check, so a
conflicting ClusterPolicy left it stale. Move the count to run right
after the GPUCluster is fetched. The finished line was written only on
normal return, so a panic in the inner reconciler left no record. Write
it from a defer with outcome "panic" and re-raise, so controller-runtime
still recovers. The ClusterPolicy start line built a logger and threw it
away; log it, so an unmatched start is visible on that path too.

Signed-off-by: Rajath Agasthya <ragasthya@nvidia.com>

19 of 19 new or added lines in 3 files covered. (100.0%)

5896 of 13066 relevant lines covered (45.12%)

0.45 hits per line