Usually my server only crashes when it's a new deployment where we've pushed bad code, or don't have an env var set up in that environment.
I would much prefer it uses a working deploy than being down

We would prefer if the user reconciles the crash loop, or set up CI rails to prevent this from happening. Our two cents.

Angelo Saraceno: from a user perspective, this is an inferior experience. When would I ever prefer a crashed server to be on prod, over a working version of the last existing server?
I don’t want the new code to be promoted to prod at all if it’s not working

Just clarifying here:

• We have a setting to infinitely restart crashed deployments

• You'll get notified when a deploy crashes

• You can setup healthchecks to assure that the deployment at least starts in the most basic case
  This should cover all basis' for the near future

We would prefer if the user reconciles the crash loop, or set up CI rails to prevent this from happening. Our two cents.

Anukriti: Hey! Please see my response above in the pinned comment section
You can prevent promotion by setting up a healthcheck (https://docs.railway.app/deploy/healthchecks)