0 0 Share PDF

How do I migrate an existing UCP cluster to the overlay 2 graph driver?

Article ID: KB000814

Issue

How do I migrate my existing Docker EE (UCP/DTR) cluster to the Overlay2 graph storage driver?

Prerequisites

  • A stable/healthy cluster. All existing unhealthy nodes should be addressed.
  • Host shell access with root permissions

Resolution

The instructions for reconfiguring each node in the cluster with the overlay2 storage driver will not cause you to lose anything in your UCP cluster configuration.

Note: It is not required but recommended to perform a backup of your EE Environment prior to performing the migration.

Note: If you are migrating from devicemapper to overlay2, the storage configured as an LVM thinpool for use with devicemapper is not automatically freed for use by overlay2. For instructions on converting the LVM thinpool to a mountable Logical Volume you can use with overlay2, please see How to convert a devicemapper thinpool volume for use with overlay2.

1. Manager Nodes

Perform the below steps one manager node at a time.

  1. In a shell session on the node follow these steps to configure to overlay2:

    https://docs.docker.com/storage/storagedriver/overlayfs-driver/#configure-docker-with-the-overlay-or-overlay2-storage-driver

  2. Log in to the UCP UI and confirm that all nodes are healthy, all managers are up, and the cluster is stable. You may notice that the node that we migrated to Overlay2 storage is unhealthy initially. Please allow a few minutes for it to become healthy before proceeding.

  3. Repeat steps 1 and 2 for the other manager nodes before proceeding to DTR Nodes.

2. DTR Nodes

Perform the below steps one DTR replica at a time.

  1. In a shell session on the node follow these steps to configure to overlay2:

    https://docs.docker.com/storage/storagedriver/overlayfs-driver/#configure-docker-with-the-overlay-or-overlay2-storage-driver

  2. Log in to the UCP UI and confirm that all nodes are healthy, all managers are up, and the cluster is stable. You may notice that the node that we migrated to Overlay2 storage is unhealthy initially. Please allow a few minutes for it to become healthy before proceeding.

  3. Repeat steps 1 and 2 for the other DTR replicas before proceeding to Worker Nodes.

3. Worker Nodes

  1. Follow the steps here to configure to overlay2:

    https://docs.docker.com/engine/userguide/storagedriver/overlayfs-driver/#configure-docker-with-the-overlay-or-overlay2-storage-driver

  2. Check the UCP UI that all nodes are healthy, all workers are up, and the cluster is stable.

  3. Repeat steps 1 and 2 for the remaining worker nodes.