Mellanox / k8s-rdma-shared-dev-plugin

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

the new kubelet --root-dir cannot be used!

sober-wang opened this issue · comments

kubelet --root-dir=/data/kubelet

rdma-share daemonset yaml file:

        volumeMounts:
          - name: device-plugin
            mountPath: /data/kubelet/device-plugins
            readOnly: false
          - name: plugins-registry
            mountPath: /data/kubelet/plugins_registry
            readOnly: false
          - name: config
            mountPath: /k8s-rdma-shared-dev-plugin
          - name: devs
            mountPath: /dev/
      volumes:
        - name: device-plugin
          hostPath:
            path: /data/kubelet/device-plugins
        - name: plugins-registry
          hostPath:
            path: /data/kubelet/plugins_registry
        - name: config
          configMap:
            name: rdma-devices
            items:
            - key: config.json
              path: config.json
        - name: devs
          hostPath:
            path: /dev/

the rdma shared plugin print logs.

2024/01/18 08:29:00 Initializing resource servers
2024/01/18 08:29:00 Resource: &{ResourceName:hca_shared_devices_a ResourcePrefix:rdma RdmaHcaMax:1000 Devices:[enp88s0] Selectors:{Vendors:[] DeviceIDs:[] Drivers:[] IfNames:[enp88s0] LinkTypes:[]}}
......
2024/01/18 08:29:00 Starting all servers...
2024/01/18 08:29:00 starting rdma/hca_shared_devices_a device plugin endpoint at: hca_shared_devices_a.sock
2024/01/18 08:29:00 Error: starting resource servers listen unix /var/lib/kubelet/device-plugins/hca_shared_devices_a.sock: bind: no such file or directory

we should make this path configurable in device plugin. currently i see its hardcoded

meanwhile you could mount the alternative kubelet path in its default place in container i.e

        volumeMounts:
          - name: device-plugin
            mountPath: /var/lib/kubelet/device-plugins
            readOnly: false
          - name: plugins-registry
            mountPath: /var/lib/kubelet/plugins_registry
            readOnly: false
          - name: config
            mountPath: /k8s-rdma-shared-dev-plugin
          - name: devs
            mountPath: /dev/
      volumes:
        - name: device-plugin
          hostPath:
            path: /data/kubelet/device-plugins
        - name: plugins-registry
          hostPath:
            path: /data/kubelet/plugins_registry
        - name: config
          configMap:
            name: rdma-devices
            items:
            - key: config.json
              path: config.json
        - name: devs
          hostPath:
            path: /dev/