error with gpushare about gpushare-scheduler-extender HOT 6 CLOSED

@cheyang done, I deploy kube-scheduler not in container, so the IP addr in the scheduler-policy-config.json file should not be '127.0.0.1', should be clusterIP.

from gpushare-scheduler-extender.

I think "aliyun.com/gpu-mem":"10G" should be changed to "aliyun.com/gpu-mem":"10" . Please take a look at https://github.com/AliyunContainerService/gpushare-scheduler-extender/blob/master/docs/userguide.md

from gpushare-scheduler-extender.

Btw, can you show me the result of kubectl describe po gpushare?

from gpushare-scheduler-extender.

@cheyang I use the "aliyun.com/gpu-mem":"2", and result is "Error: failed to start container "test": Error response from daemon: OCI runtime create failed: container_linux.go:344: starting container process caused "process_linux.go:424: container init caused "process_linux.go:407: running prestart hook 0 caused \"error running hook: exit status 1, stdout: , stderr: exec command: [/usr/bin/nvidia-container-cli --load-kmods configure --ldconfig=@/sbin/ldconfig --device=no-gpu-has-2MiB-to-run --utility --pid=14123 /home/docker/overlay2/b20cfd4438c42624b7b077786963b307545df9e2e205687291a3f5436908b306/merged]\\nnvidia-container-cli: device error: unknown device id: no-gpu-has-2MiB-to-run\\n\""": unknown"

from gpushare-scheduler-extender.

@cheyang my kubernetes version is v1.10.6

from gpushare-scheduler-extender.

Please run kubectl-inspect-gpushare first to check the gpu memory unit you are using. Looks like it's MiB.

You can set "aliyun.com/gpu-mem":"1024" to try.

from gpushare-scheduler-extender.