Skip to content

Latest commit

 

History

History
45 lines (32 loc) · 1.88 KB

update_quobyte_csi_or_clients.md

File metadata and controls

45 lines (32 loc) · 1.88 KB

Index

Update Quobyte CSI Driver

  1. Remove existing CSI Driver

      helm delete <CSI-DRIVER-NAME>
  2. Install new CSI Driver

Impact: Removing and reinstalling Quobyte CSI driver should not disrupt application pods with already mounted volumes. Only, new dynamic volume provisioning, volume mount, delete requests will fail temporarily until the new CSI Driver is available.

Update Quobyte Client

  1. Upgrade Quobyte client on single k8s node and wait until all application pods are recreated and running on that k8s node.

  2. Once the application pods are running on the node with upgraded Quobyte client, proceed to next k8s node and repeat step 1.

Impact: On the particular k8s node where Quobyte client is updated, all application pods with Quobyte Volumes will be disrupted due to the stale mount points. Application pod termination, in such case is handled by our CSI sidecar. See Application Pod Recovery for more information.

Application Pod Recovery

A crashing Quobyte client/upgraded Quobyte client leaves all the existing application pods with Quobyte CSI volumes in invalid and stale state on that particular k8s node where Quobyte client is upgraded or crashed. To recover, existing application pod must be deleted and a new pod should be created. The pod monitoring sidecar container shipped with Quobyte CSI automatically deletes the application pods with stale Quobyte CSI volumes and leaves the creation of new application pod to k8s.

For k8s to schedule a new pod in place of the killed pod with stale Quobyte mounts, your application should be deployed as Deployment/Statefulsets/Replicaset but not as a plain Pod.