A Friday outage that should not have happened A commerce team had done what most architecture checklists recommend. Their API ran in two regions, database replicas were healthy, autoscaling worked, and traffic failover tests passed monthly. On a Friday release,…
Author: Ankur Sharma
-
The Build Script That Touched Production Secrets: A 2026 Node.js Permission Model Rollout Playbook
A practical Node.js permission model rollout guide: least-privilege runtime, safer npm script handling, and incident-tested steps for production teams.
-
The UI Felt Fine in QA, Then Collapsed at Scale: A 2026 Frontend Performance Playbook for Real-World Interaction Integrity
A launch story with no outage and plenty of user pain A consumer app team shipped a redesigned onboarding journey on Friday evening. It looked polished, load times were acceptable, and all synthetic checks passed. By Saturday afternoon, support volume…
-
Your App Is Crashing, But the Store Review Takes Hours: A 2026 Mobile Kill-Switch Playbook
Learn a practical mobile app kill switch architecture using Firebase Remote Config, staged rollouts, and safe fallback paths to recover from bad releases fast.
-
The Healthy Cluster, Unhealthy System: A 2026 Backend Reliability Playbook for Drift, Sabotage Resistance, and Fast Recovery
A Saturday incident where everything looked “up” A logistics startup had a normal weekend traffic spike. Kubernetes was healthy, CPU looked good, and error rates stayed low. Yet customer complaints surged. Delivery slots vanished, then reappeared. Some orders were marked…
-
From Pager to Proof: A 2026 Java Runbook for Continuous JFR Capture and Fast CPU Triage
Java Flight Recorder in production: set up continuous JFR capture, extract the right incident window, and triage CPU spikes fast with safer profiling tradeoffs.
-
The Benchmark Passed, Production Regressed: A 2026 AI/ML Playbook for Durable Model Operations
A launch story with great metrics and bad outcomes A product team shipped a new support assistant after excellent offline evaluation. Their benchmark score improved, latency looked acceptable, and cost per request dropped. In week one, executives were happy. In…
-
The IAM Trust Policy That Didn’t Scale: A 2026 Migration Playbook from IRSA to EKS Pod Identity
Practical 2026 guide to EKS Pod Identity migration from IRSA, with safe rollout steps, IAM tradeoffs, troubleshooting, and multi-cluster validation checks.
-
The Legacy Plugin That Almost Took the Community Offline: A 2026 WordPress Engineering Playbook for Safe Extensibility
A quick story from a weekend incident nobody expected A media team running a revived legacy social site had a normal Saturday deployment: one plugin update, one theme tweak, and a small change to user profile caching. Traffic looked healthy…
-
The Offboarding Incident: Replacing Fragile PAT Scripts with GitHub App Installation Tokens
Replace brittle PAT scripts with GitHub App installation tokens: least-privilege permissions, short-lived creds, and automation that survives offboarding.
-
The Dashboard Said “All Good” While Data Was Wrong: A 2026 SQL Reliability Playbook for Human-Verified Analytics
A real incident where automation passed and judgment didn’t A growth team shipped a pricing experiment and monitored it through a near-real-time dashboard. Their SQL pipeline was fully automated, tests were green, and anomaly alerts stayed quiet. By the third…
-
The PWA Was Fast Until Monday Morning: A 2026 Web Development Playbook for Navigation Preload and Safe Service Worker Updates
Fix intermittent PWA slowdowns using service worker navigation preload, safe cache headers, and controlled updates that prevent post-deploy blank screens.