Platform alias routes on tito 404: empty alias rows never resolve an upstream Issue open
Migrated from github.com/carlosframework/platform issue #148, opened 2026-08-11. GitHub issues are closed for this repo; see AGENTS.md, "Work is tracked on amadan, not GitHub".
Still outstanding: Platform alias routes for bespoke-unit apps 404 because empty alias rows never resolve an upstream.
---
Found during the PR #144 rollout (2026-08-11). Both platform-minted aliases for titogo on carlos-tito-ie 404 at the edge with no X-Carlos-Version:
- titogo.bae.platform.tito.io — the app's real account; should serve the app's apex upstream (go.tito.io), returns 404.
- titogo.bab.platform.tito.io — stale twin from before the bab→bae move; reconcile keeps re-minting it because apps/bab/titogo/ still holds channels/releases/config.json in the bucket (separate cleanup, operator decision).
Registry rows for both: kind=instance, no channel, no version, no backing, no unit — bare rows nothing serves. The alias→upstream resolution that works for exec/carlos-app-backed apps on flagship (hello.bdf.oncarlos.com) appears not to follow tito's bespoke-unit-backed apex routes.
Per the no-infra rule: an alias the platform mints must serve (or not exist) — a member can see these hosts on their app page and gets an unexplained 404.
🤖 Generated with Claude Code