Create device-plugin-ds.yaml, Daemon don't create device-plugin-ds pod

AliyunContainerService / gpushare-scheduler-extender

GPU Sharing Scheduler for Kubernetes Cluster

Apache License 2.0

1.39k stars 308 forks source link

Create device-plugin-ds.yaml, Daemon don't create device-plugin-ds pod #176

Closed echo3215987 closed 2 years ago

echo3215987 commented 2 years ago

When I execute cmd "kubectl create -f device-plugin-ds.yaml", Daemon don't create device-plugin-ds pod

list all of kube-system pod
list /var/lib/kubelet/device-plugins, no aliyungpushare.sock file
k8s version(1.23.0)

If anyone have the same issue and fix it ? Could give me something clue? Thanks.

echo3215987 commented 2 years ago

I fixed the issue. When I create a job for test without gpu. The key sentence of the describe pod info is default-scheduler 0/1 nodes are available: 1 node(s) had taint {node-role.kubernetes.io/master: }, that the pod didn't tolerate.

detail information as below.

So I execute: kubectl taint nodes tpeaa03-precision-7920-tower node -role.kubernetes.io/master- It works and shows gpushare-device-plugin. gpushare-device-plugin-ds-hm52c