Consistency, Validation, and Repair

After incidents, upgrades, or partial failures, you need a reliable way to prove and restore consistency.

Repair Workflow

1. Run rclone check to identify mismatches.
2. Re-sync affected scope.
3. Re-run validation and archive logs.

repair-loop.sh

rclone check /srv/data remote-prod:backup/data --one-way --combined /tmp/rclone-check.txt
rclone sync /srv/data remote-prod:backup/data --progress
rclone check /srv/data remote-prod:backup/data --one-way

Validation Modes

Mode	Use when
Size/time compare	Backend hash support is limited
Hash compare	High-integrity validation needed
Partial scope validation	Very large datasets

tip

For large estates, validate by dataset domain (db/, media/, configs/) to shorten mean-time-to-repair.

Common Pitfalls

Pitfall	Consequence	Prevention
Re-syncing everything first	Longer outage window	Check first, then targeted repair
No evidence logs	Hard postmortem	Keep combined check output files
Single-threaded repair on huge sets	Slow recovery	Tune transfers/checkers carefully

What's Next

• Core Commands Cheatsheet
• Safety Checklist Cheatsheet