Disaster Recovery for Hosted Control Planes

This section covers all backup, restore, and disaster recovery procedures for HostedClusters managed by HyperShift. It is organized around scenarios (what you want to achieve), methods (how to achieve it), and platform-specific guides (caveats per provider).

Choosing the Right Approach

Use the following decision tree to find the right procedure for your situation:

flowchart TD
    A[What happened?] --> B{Is the Management\nCluster recoverable?}
    B -->|Yes| C{What failed?}
    B -->|No| D[Cross-cluster migration]

    C -->|Etcd member lost| E[Etcd Recovery\n- automatic or manual]
    C -->|Control plane down\nor data corruption| F[Same-cluster restore]
    C -->|Accidental HC deletion| F

    D --> G{Do you have\nfixed hostnames?}
    G -->|Yes| H[Cross-cluster migration\nprocedure]
    G -->|No| I[⛔ Cannot migrate\nConfigure fixed hostnames first]

    F --> J{Which backup\nmethod?}
    J -->|OADP with volume snapshots| K[OADP Restore]
    J -->|Etcd snapshot - Tech Preview| L[Etcd Snapshot Restore]
    J -->|Manual etcd backup| M[Manual Restore]

    click E "etcd-recovery.md"
    click F "scenarios/same-cluster-restore.md"
    click H "scenarios/cross-cluster-migration.md"
    click K "methods/oadp.md"
    click L "methods/etcd-snapshot/"
    click M "methods/manual-etcd-backup.md"

Supportability Matrix

Platform	OADP Volume Snapshot	Etcd Snapshot (TP)	Same-cluster Restore	Cross-cluster Migration	Node Readoption
AWS	✅ Supported	✅ Tech Preview	✅ Supported	⚠️ Procedure documented — not yet supported (no E2E)	❌ Not supported
Azure	✅ Supported	✅ Tech Preview	✅ Supported	⚠️ Procedure documented — not yet supported (no E2E)	❌ Not supported
Agent / Bare Metal	✅ Supported	✅ Tech Preview	✅ Supported	⚠️ Procedure documented — not yet supported (no E2E)	✅ OCP 4.19+ / MCE 2.9+
KubeVirt	✅ Supported	⚠️ Not validated	✅ Supported	⚠️ Procedure documented — not yet supported (no E2E)	❌ Not supported
OpenStack	⚠️ Tech Preview	⚠️ Not validated	⚠️ Tech Preview	⚠️ Procedure documented — not yet supported (no E2E)	❌ Not supported

Cross-cluster Migration Support Status

Cross-management-cluster migration procedures are documented in this guide but are not yet officially supported. End-to-end test coverage for cross-cluster scenarios does not exist yet. Use these procedures at your own risk in non-production environments, or as a last-resort disaster recovery measure.

Documentation Structure

Scenarios

Step-by-step guides organized by what you want to achieve:

Same-cluster Restore: Restore a HostedCluster on the same Management cluster where the backup was taken.
Cross-cluster Migration: Migrate a HostedCluster to a different Management cluster (unsupported — procedure documented for reference).

Backup and Restore Methods

Detailed reference for each backup/restore mechanism:

OADP (OpenShift API for Data Protection): The primary backup/restore method using Velero and the OADP HyperShift plugin.
Etcd Snapshot Backup (Tech Preview): Alternative method using native etcd snapshots instead of volume snapshots.
Manual Etcd Backup: Fully manual etcd snapshot and restore process (requires API downtime).
DR CLI Commands: HyperShift CLI commands for creating OADP backups and restores.

Platform Guides

Provider-specific configuration, caveats, and examples:

AWS: OIDC fixup, ExternalDNS cleanup, endpoint access considerations.
Azure: Workload Identity configuration, Azure Blob Storage setup for etcd snapshots.
Agent / Bare Metal: InfraEnv lifecycle, Assisted Installer database, node readoption.
KubeVirt: VM recreation, boot image PVC filtering.
OpenStack: CSI driver considerations, floating IP pools.

Reference

Prerequisites: Requirements for all DR operations including service publishing strategy.
Etcd Recovery: Manual etcd member recovery (operational procedure, not full DR).
Troubleshooting: Common issues and their resolutions.