Navigating Disaster Recovery in Kubernetes and Crossplane: Strategies and Tools for Managing Stateful Workloads with GitOps for Platform Teams
summary
Claim: GitOps alone doesn't solve disaster recovery for stateful cloud resources. Concrete tools and patterns: Velero backups to cross-region S3 for CRDs and PVs, Crossplane observe-only / orphan-delete modes to avoid accidental data deletion, and S3 bucket replication plus cloud provider DB snapshots for RPO/RTO tradeoffs. Practical pitfalls: provider CRD ownership and upgrade rollbacks can block recovery, externalName/UID mismatches cause duplicate resources, and fastest recovery often means manual restore + webhook mutations to switch regions before reattaching full GitOps automation.
AI-written summary · about
chapters (11)
- 0:00Intro
- 1:52Platform Engineering
- 2:36GetUPS
- 8:57Resilience
- 10:42Disaster Recovery
- 12:32Recovery Objectives
- 13:40Recovery Spectrum
- 15:28Crossplane
- 21:15Crossplane on
- 23:37Crossplane issues
- 30:55Summary
related talks
30mIs Your Kubernetes Disaster Recovery Actually Ready?
17mUsing GitOps for Both Infrastructure and Applications
30mGitOpsifying Cloud Infra with Crossplane, ArgoCD
29mPutting the Ops in GitOps - New Challenges in Managing Cloud Native Environments
35mDisaster Recovery for your Kubernetes Clusters
33mPlan to Fail: A Good Captain Doesn’t Sail Without Life Rafts
34mKubernetes Backup and Migration Strategies with Velero
46mKubernetes Stateful Application Disaster Recovery with Zero Data Loss