Department 8
Ops and infrastructure
- 8.1Runbook indexEvery runbook and when to use itEvery runbook in one list, so nobody runs a script from memoryDraft
- 8.2Deploy and promoteStep by step, checks, rollbackStaging takes every tested change; production only by promoteDraft
- 8.3RestoreBackups, restore steps, the restore testBackups exist for every site, so a restore is a routineDraft
- 8.4Wipe data and reset jamsWhat each wipes, what it never touchesThree ways to start over, and none of them touch connectionsDraft
- 8.5DNS and domainsEvery domain, where it points, the Mac DNS lessonOne load balancer, a few Cloudflare records, a two minute checkDraft
- 8.6Secrets and keysWhich secret, where, who rotates itEvery key sits in Secret Manager; nobody owns rotation yetDraft
- 8.7Monitoring and alertsUptime checks, alert policies, who gets themTwo layers watch Brew and email you; nobody is on callDraft
- 8.8IncidentsCalling one, roles, the postmortem templateNo incident process yet; here is one to acceptNeeds you
- 8.9Costs and quotasCloud, AI, data, credits, monthly spendCaps are set in code, but no monthly spend figure exists anywhereNeeds you
- 8.10Protected connectionsArata's Meta and Shopify, the guard, never disconnectArata's Meta and Shopify are never disconnected, and code refuses to tryDraft
- 8.11Background jobsWatchers, cadences, leases, spend capsOne worker, one lease, and every timer runs on the lease holderDraft
- 8.12Hosting mapEvery site and service: product, demo, docs, handbookEverything runs in one Google Cloud project, with Cloudflare handling DNSDraft