render_setup

Render Setup (superseded)

Status: no longer in use as of 29 July 2026.

Render is not part of Mealchemy's deployment. The backend runs on Google Cloud Run, deployed by the CD workflow. See Deployment for the current architecture.

This page is kept for historical context only. Nothing here should be treated as the live system.

---

Why it was retired

Render hosted the Spring Boot backend during sprint 1, purely so the app was reachable over the internet and Uptime Robot had a URL to ping. It was always a stop-gap while the real deployment was being built.

It was replaced because:

• The free instance is not permanent. Free Render Postgres instances expire after a fixed window, and the free web service spins down after 15 minutes of inactivity.
• Cloud Run gives per-revision traffic control. That is what makes the rollback workflow possible: rolling back is a traffic shift to an existing revision, not a rebuild.
• Secrets, identity, and deployment are separated on GCP. Secret Manager holds credentials, Workload Identity Federation removes the need for a stored deploy key, and three separate service accounts split the deploy identity from the two runtime identities.

The full comparison of alternatives is in section 12 of Deployment.

What replaced it

Was	Now
https://mealchemy-backend.onrender.com	https://mealchemy-backend-prod-otygypdv7a-ey.a.run.app (production)
	https://mealchemy-backend-staging-otygypdv7a-ey.a.run.app (staging)
Render auto-deploy on push to main	CD workflow on push to main (production) or deployment (staging)
Render-managed Postgres	Neon Postgres 16, staging and production branches

The health endpoint contract is unchanged:

{ "status": "UP" }

Uptime Robot now monitors the Cloud Run production health endpoint as a keyword monitor, which asserts the response body actually reports UP rather than only checking that the request returned HTTP 200. This powers the uptime badge on the repository README.

Historical note on the monitoring gap

The Render service stopped responding on 20 June 2026. Uptime monitoring was repointed at Cloud Run on 29 July 2026, which is the start of the current monitoring window.