Sleeping with one eye open: production support
summary
Covers concrete failures and fixes: a bank migration locks 1.9 million customers out for a month, a certificate renewal knocks 30 million users offline, Google SSO breaks Gmail/YouTube for an hour, and a sudden Signal surge exposes capacity limits. Recommends two-person roles (monitor + communicator), runbooks with architecture diagrams and repo/status links, alerting baselines, unique-user impact counts, a buddy on-call rota, and simple diagnostics built (network-report tool for IPv6 issues) that actually find and fix root causes.
AI-written summary · about
chapters (11)
- 0:00Production support intro
- 3:26Why support matters
- 8:43First 3am steps
- 12:58Roles and questions
- 18:49Essential support tools
- 26:15Communication in crises
- 30:10Bypassing processes risks
- 34:23Downtime and self-care
- 36:33Incident wash-ups
- 39:00Release safety strategies
- 40:24Support schedule management
related talks
26mPaid to Sleep
15mMaintaining reliable systems
39mOops. I broke the Google, now what?
45mLet's play a real life Dungeon and Dragons
52mPersistent SRE Antipatterns: Pitfalls On the...
52mPerformance & Stability Anti patterns
01h 03mWhispers in the Chaos: Monitoring Weak Signals- 01h 03mWhispers in the Chaos: Monitoring Weak Signals