edgegap-launch

Topics in the source

• Launching and operating on Edgegap
• Pre-launch checklist
• Launch-day behavior
• Observability
• Long-running and persistent servers
• When something breaks
• After launch

This assumes a test deployment already accepted a real client (edgegap-local-testing). This skill is the gap between "it works" and "it's operable". When you act on the account during a launch, follow the edgegap skill's references/agent-rules.md: this is real players and real billing. Account and tier Free tier limits are not launch limits: one deployment at a time, 60-minute deployments, 2 apps, 2 versions, 10 GB registry, 1.5 vCPU / 3 GB, matchmaker stops after 3 hours. Upgrade to Pay as You Go before any public test. It also unlocks image caching and detailed metrics. Move the matchmaker off the free tier (it stops 3 hours after each start) to a private cluster tier sized for launch traffic (Hobbyist, Studio or Enterprise; they differ in rate limit and redundancy). Pricing: https://edgegap.com/resources/pricing#matchmaker. Tell Edgegap about the launch date and expected traffic in advance (support ticket, Discord, or info@edgegap.com). They plan regional capacity and can raise rate limits. Tokens The Edgegap API token is organization-wide. Only the backend and CI hold it, from a secret store. None in clients, server images, deployment env vars or repos. Use a separate token per

This is an excerpt. Refer to the complete source and its current requirements.