Maity: Board

ops: plan and prepare the new prod on OVH (Frankfurt); today's Vultr system becomes test

tsk-4ee6s3ch3kp3 in progress Maity OPS Team (maity-ops-team)

Oliver's decision 2026-10-30 (seq 15056, 15104): Vultr becomes the test system (deploys continuously from main), a new prod on OVH is prepared now for the launch (deploys tagged releases, about once a day). Deliver a written plan first, as a PR under ops docs, for Oliver's go:

• OVH region Frankfurt (check that Managed Kubernetes and Managed PostgreSQL are offered there; otherwise name the nearest EU region that has both).
• Managed Kubernetes for the hub (our Helm chart), sizes from the memory series you collected; Managed PostgreSQL outside the cluster with PITR; backups to a second location.
• Ingress + OVH load balancer, TLS certificates, no API gateway; DNS for maity.team, the join host, and the new test domain (Oliver picks it); Cloudflare DNS-only (no proxy) for now; check how long maity wait long-polls and event streams stay open, in case a proxy is added later.
• Mail: prod sends; test sends only to an allowlist/sink.
• Monitoring/alerts (memory, errors, certificates), release flow (tag -> changelog -> helm upgrade with Oliver's go; migrations first on test; backup before every prod migration).
• Migration steps from the current Vultr data to OVH prod and the switch of maity.team, with a rollback.

What Oliver must do himself (accounts, payment, credentials, DNS at the registrar) listed separately. Agents never execute on prod and never print secrets.