For DevOps and operators#

Your questions#

Question	Answer and where
Who do I need, and how much work is it?	Skills and effort
How do I install it?	Creating workspaces: scripts/provision-workspace.sh, TLS, first administrator
What is the order of things?	Operations: start here
What does each variable do?	Configuration reference
Which secrets exist and how do I rotate them?	Secrets and keys
What is encrypted, and how do backups work?	Encryption, backup, restore and DR
How do I upgrade, and roll back?	Upgrade and migrations; migrations run when core-api starts
What do I monitor?	Observability and logging: readiness (/health) and liveness (/health/live), logs, and Prometheus metrics that stay off until METRICS_TOKEN is set; nothing sends alerts
How much load can it take?	Capacity and scaling (one load test on a development machine, 2026-10-18; yours will differ)
How do I run shared rate limits across gateway processes?	Optional Valkey: API clients, section 9
How do I create a demo instance?	Public sandbox
Which object stores work?	Object storage

Runbooks: what has been run, and what is yours to rehearse#

Run by scripts on a throwaway workspace: the restore of a destroyed workspace (scripts/restore/rehearse.sh, last passed 2026-10-20) and the upgrade from the previous release (scripts/upgrade/upgrade-test.sh, v0.1.1 to v0.2.0). The upgrade of the public host was done by hand on 2026-10-19. What no one has run for you: a restore on your host with your data, and your own timings. Rehearse each once and record the date.