Renaud Gaubert 170bae8a5d Return a NoopManager if metricset does not container the accelerator value ... This is because the NVIDIA manager opens a handle on nvml at NewNvidiaManager time. This is problematic in a kubernetes setting where kubelet now has an open handle on the NVIDIA driver, preventing an update of the NVIDIA driver unless kubelet is restarted. Additionally with the new metrics pipeline in Kubernetes, metrics are now expected to be collected through a container rather than through the kubelet itself. Signed-off-by: Renaud Gaubert <rgaubert@nvidia.com>		2020-06-05 00:06:48 +00:00
..
container_test.go	golanci-lint based cleanup (#2503)	2020-04-20 10:53:19 -07:00
container.go	Add psr option when ps containers	2020-05-27 15:01:54 +08:00
manager_test.go	add unit test cases for Getting subcontainers info if scans with more subcontainers	2020-05-15 11:49:05 +08:00
manager.go	Return a NoopManager if metricset does not container the accelerator value	2020-06-05 00:06:48 +00:00