backup-dr

Back up and restore Kubernetes workloads with Velero, and prove recovery works. Covers Velero on EKS via Flux (IRSA, S3 backup location, EBS snapshots), schedules and retention, what a backup does not

SKILL.md

GitHub

A backup is only as good as the last time it was restored. Set up Velero, then prove each namespace comes back and measure how long it takes.

Read references/backup-dr.md before responding. It holds what Velero does and does not capture (and how GitOps changes that), configuration decisions, the restore drill, cross-cluster DR, rollback and troubleshooting. The working files are in examples/backup-dr.

This skill is interactive by default. With no arguments, ask what must be recoverable (namespaces, volumes, the whole cluster), the recovery point and recovery time objectives, and whether the target is the same cluster or a new one, then run schedule or dr. Never run a restore, delete a namespace, or patch a backup location yourself: produce the commands and let the user run them.

Mode dispatch

Mode What it does
install The protected backup bucket from examples/backup-dr/bucket (versioning, noncurrent-version retention, a bucket policy that lets only break-glass roles delete versions or lift the protection), then Velero via Flux from examples/backup-dr with IRSA and the scoped IAM policy
schedule Schedules and retention derived from the recovery point objective; which namespaces need more frequent backups
drill Run and interpret restore-drill.sh; compare the measured time with the recovery time objective
restore Restore a namespace or specific resources, including into a different namespace
dr Recovery into a new cluster: Flux bootstrap first, read-only backup location, then application namespaces
debug Failed or partial backups and restores, from velero backup/restore describe --details and logs

Output format

Every mode ends with:

  • Changes or commands, exact
  • Blast radius, in one line
  • Validation, as commands (a drill counts)
  • Rollback