Dev.to
7/13/2026

4 self-hosting failures that silently report success while broken
Original: 4 self-hosting failures that return success
Short summary
A self-hoster shares four costly failures where services reported success while being broken: Sablier serving 200 OK loading pages to WebDAV clients, Docker healthchecks silently failing after curl was removed from minimal images, empty database dumps reporting success, and MariaDB accepting connections before being truly ready. Each failure cost an evening to diagnose. Fixes include using bash /dev/tcp for healthchecks and verifying backup file sizes, not just exit codes.
- •Sablier returns 200 OK loading pages that break WebDAV sync
- •Minimal Docker images drop curl, silently breaking healthchecks — use /dev/tcp instead
- •Empty database dumps and premature MariaDB readiness also report false success
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



