Backup and Restore Runbook

HeliosDB Backup and Restore Runbook

Use this runbook for backup verification, point-in-time recovery, full restores, and disaster recovery testing.

Backup Verification

Verify backups before you need them:

1. Confirm the latest full backup completed.
2. Confirm incremental or WAL backups are continuous.
3. Verify checksums and manifests.
4. Run a scheduled restore test in an isolated environment.
5. Compare row counts or application-level validation queries.

Example checks:

heliosdb-backup list --last 7days
heliosdb-backup verify --backup-id latest --checksum
heliosdb-backup verify --backup-id latest --completeness

Point-in-Time Recovery

Use point-in-time recovery when data must be restored to a specific timestamp or LSN.

1. Identify the recovery target time.
2. Stop writes to the affected database if corruption is spreading.
3. Restore the most recent full backup before the target time.
4. Replay WAL to the target.
5. Start the restored database in isolation.
6. Validate data before promoting or exporting.

heliosdb-backup restore \
  --backup-id latest-before-target \
  --target-time "YYYY-MM-DDTHH:MM:SSZ" \
  --target restore-cluster

Full Restore Procedure

Use a full restore for complete environment loss or planned migration validation.

1. Provision clean storage and compute.
2. Restore the selected full backup.
3. Apply incremental backups or WAL if required.
4. Start HeliosDB in recovery mode.
5. Validate schema, row counts, and application smoke tests.
6. Switch traffic only after validation succeeds.

Cross-Region Restore

For regional recovery:

1. Confirm backup availability in the destination region.
2. Restore into an isolated cluster.
3. Validate network, secrets, and endpoint configuration.
4. Run application smoke tests from the destination region.
5. Update routing after approval.

Recovery Time Estimation

Estimate recovery time from:

• Backup size and storage throughput.
• WAL volume since the selected full backup.
• Network transfer time for cross-region restores.
• Validation query duration.
• Application cutover and DNS or load balancer timing.

Backup Troubleshooting

Symptom	Check	Action
Missing WAL segment	Backup chain continuity	Restore from earlier full backup or alternate replica
Checksum failure	Storage integrity and manifest	Do not use the backup; select a verified backup
Restore is slow	Network and disk throughput	Restore in-region or increase temporary capacity
Restored DB will not start	Logs and control metadata	Validate backup completeness and configuration