Hi everyone, I really appreciate any input here, we're looking to migrate away from Vercel due to constraints getting in the way of our application build on Claude using API calls and integrated with Supabase (and currently Vercel).

The most important questions are #1 and #3 below, if anyone knows, much appreciated.

• 1. Background process limits — Is there a time limit on background tasks/processes running outside of an HTTP request? If my Node.js server kicks off a 20-minute async task after responding to an HTTP request, will it be interrupted?

• 2. Memory limits — What's the memory limit per container? My app makes multiple large AI API calls with responses of 25K+ tokens. (I want at least 512MB, ideally 1GB+)

• 3. Sleep/inactivity — Does Railway sleep or shut down my app after a period of no incoming requests? I need a persistent always-on server, not a scale-to-zero model.

• 4. Deploy downtime — When I push a new deploy, is there zero-downtime deployment or does the app restart? If a pipeline is mid-run during a deploy, what happens?

• 5. Custom domains + SSL — Can I bring my own domain with automatic SSL?

• 6. Pricing model — What's the actual cost for a Node.js app running 24/7 with occasional spikes of 15-20 minutes of CPU-intensive work?

• 7. Egress/bandwidth — Are there egress limits? My app makes multiple external API calls (Anthropic, SerpAPI, Supabase) per run.