Skip to content

fix(web): stop health checks restarting healthy relays - #432

Merged
sleep3r merged 1 commit into
mainfrom
codex/fix-web-health-431
Sep 30, 2026
Merged

sleep3r merged 1 commit into
mainfrom
codex/fix-web-health-431

Conversation

@sleep3r

@sleep3r sleep3r commented Sep 30, 2026

Copy link
Copy Markdown
Owner

WEB health checks restarted healthy relays when [web].listen selected a non-loopback address or the public root returned its intended 404 without public_dir. Read both listener aliases in configuration order, accept completed local HTTP responses, and allow public HTTPS 404 while continuing to report other 4xx/5xx and transport/TLS failures without restarting the relay.

Add executable shell regression coverage with real curl HTTP fixtures and stubbed systemd, wire it into CI, and document the follow-up sudo mtbuddy update --force required to replace embedded health scripts after upgrading an older updater.

Validation: 495 Zig tests passed in Linux with Zig 0.16.0; all 20 Python regression tests passed; Linux x86_64_v3+aes release cross-build and formatting passed. The new tests reproduced the original failures before the fix. Independent review found no blockers.

Fixes #431.

@sleep3r
sleep3r merged commit abe7ad4 into main Sep 30, 2026
10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Bug]: mtproto-web-health.sh restarts a healthy WEB relay every minute (reads [web].host instead of listen; treats the v1.15.3 404 on / as failure)

1 participant