Troubleshooting

Troubleshooting

Common errors

The CSI driver failed to bind a PVC to a PV

You can identify this error in the following ways:

• When checking the status of a PVC using kubectl get pvc -n <px-namespace>, the PVC displays as in the pending state.

• Logs contain the following message:

  error level=error msg="Could not init boot manager" error="failed to generate a new node identity: create CSI volume (CreateVolume) error: PVC: <px-namespace>/px-do-not-delete-... failed to get bound to a PV"

  To correct this error, check if the CSI driver exists in your cluster by running the following command:

  kubectl get csidriver

  NAME                     ATTACHREQUIRED   PODINFOONMOUNT   MODES        AGE
  csi.vsphere.vmware.com   true             false            Persistent   6d22h

• If the driver is not present in the system, contact your cluster administrator to install the vSphere CSI driver.

• If the CSI driver exists, check the StorageClass name you’re using for installation. Ensure that StorageClass has the CSI driver name set as a provisioner.

Collect logs

• Collect portworx pods logs:

  kubectl logs --since=0 -n <px-namespace> -l name=portworx

• Find the vsphere-CSI-controller-*** pod in your system:

  kubectl get pods --all-namespaces | grep vsphere-csi-controller

  Once found, collect CSI driver logs:

  kubectl logs --since=0 -n <px-namespace> <vsphere-csi-controller-pod> csi-attacher > csi_attacher.log
  
  kubectl logs --since=0 -n <px-namespace> <vsphere-csi-controller-pod> vsphere-csi-controller > vsphere-csi-controller.log
  
  kubectl logs --since=0 -n <px-namespace> <vsphere-csi-controller-pod> csi-provisioner > csi-provisioner.log

• Collect PVC, PV and VolumeAttachments lists:

  kubectl get pvc -n <px-namespace> > pvc.log
  
  kubectl get pv > pv.log
  
  kubectl get volumeattachment > volumeattachment.log