Using Velero to back up and restore Kubernetes cluster resources
Velero is an open source tool that can be used with CloudBees CI on modern cloud platforms to back up and restore Kubernetes cluster resources.
Using Velero for backup and restore is optional. The information provided on this page provides tips for using Velero with CloudBees CI on modern cloud platforms, but does not replace Velero’s documentation.
Follow the Velero installation guide to install the Velero CLI on a machine that has access to the Kubernetes cluster.
Follow the setup instructions for your storage provider to install the cluster components for Velero. Although Velero can be installed without a storage plugin, it cannot back up CloudBees CI on modern cloud platforms until the storage is integrated.
Keep the following recommendations in mind while working with Velero:
When creating a backup and a schedule, include
For backing up the cluster-scoped resources, use
When creating a restore, use
During a restore, resources that already exist are not overwritten, so you can’t restore into a Kubernetes cluster on which CloudBees CI is already installed. You must restore into an empty cluster.
For Velero version 1.5, restores must be performed in the same availability zone and region in which the backup was created.
The following examples show how the CloudBees CI on modern cloud platforms engineering team uses Velero. These are for example purposes only and may need to be adjusted to fit your environment. CloudBees does not guarantee that these examples will work for your environment. Refer to the Velero documentation for specific instructions.
velero schedule create daily --schedule=”@every 24h” --snapshot-volumes --include-cluster-resources
The default time-to-live of 720 hours was used. CloudBees CI on modern cloud platforms backups typically have between 1800 and 2000 objects and take approximately 2 minutes to complete.
velero schedule create cluster-backup --snapshot-volumes --include-cluster-resources
velero restore create --from-backup cluster-backup-<timestamp> --restore-volumes
Testing a backup and restore of a single master in GKE with a 100GB PVC that had 51GB used, the backups consistently took 7-8 seconds and the restore completed in under 1 minute.