Disaster Recovery Guide

Objective

Restore read availability and consistency from S3/MinIO manifests and chunks with minimal time-to-first-read.

Recovery Workflow

1. Provision replacement Astra nodes.
2. Configure object-tier settings (ASTRAD_S3_ENDPOINT, ASTRAD_S3_BUCKET, ASTRAD_S3_PREFIX).
3. Start nodes with clean local data directories.
4. Verify manifest/chunk retrieval and revision restoration.
5. Re-enable write traffic after quorum and key-parity checks.

Verification Commands

etcdctl --endpoints=http://127.0.0.1:2379 endpoint status -w table
etcdctl --endpoints=http://127.0.0.1:2379 get /__tenant/ --prefix --keys-only

Failure Modes to Watch

• Missing object chunks.
• Checksum mismatch.
• Partial node restore with divergent revisions.