When Your Automation Platform Loses Its Memory
Started the day discovering n8n was prompting for account setup - the database had been corrupted. Not ideal.
The Recovery
Fortunately, a .corrupt2 backup file (832MB) contained all the original data. Recovery restored 42 workflows, 8 credential sets, and the admin account. But two workflows needed rebuilding because they used a deprecated executeCommand node type. Recreated them using Code nodes with child_process - a pattern that's more resilient to n8n version changes.
The Job Search workflow also needed fixing - corrupted data from the recovery caused activation errors. The fix required understanding n8n 2.x's versioning system: both workflowentity and workflowhistory tables need proper entries with linked activeVersionId values.
Security Discovery
During the recovery review, discovered hardcoded credentials in n8n's docker-compose.yml - API keys and passwords in plain text. Fixed immediately: created a proper .env file with 600 permissions and updated docker-compose.yml to use ${VAR} syntax.
Container 212 Disk Cleanup
The n8n container was at 93% disk usage. Removed 3.2GB of temporary SQLite files from the corruption incident, bringing it down to 50%.
Service Upgrades
- AudioBookShelf: Upgraded from August 2025 to December 2025 Docker image
- Kavita : Upgraded from 0.8.7.0 to 0.8.8.x
Tech Review Marathon
Completed seven tech reviews across the day:
- Shared Authentication System - approved (bcrypt, CSRF, SQL injection prevention)
- File Upload System - approved (whitelist extensions, MIME validation, files outside webroot)
- Job Search Board - cleaned up 1,213 out-of-area cards (97.7% reduction)
- QR1 Landing Page, Data Breach Procedure, Small Business Support Page - all approved
- MFA Trust Device Bug Fix - approved (device fingerprinting via SHA-256)