A Cost-Anomaly Job Can Run Green and Check Nothing: CUR Lag and a date == today Gate
Our daily anomaly sweep ran on lagged cost data and only evaluated rows dated today, so it checked zero services with no error. The bug and the fix.
6 min read
Founder notes
What I learn building CloudWise as a solo founder: SR&ED, build-in-public write-ups, bugs and the numbers behind them. Also as a podcast
7 posts tagged "Engineering"
Our daily anomaly sweep ran on lagged cost data and only evaluated rows dated today, so it checked zero services with no error. The bug and the fix.
6 min read
A green unit test proves your code does what you think AWS does, not that AWS does it. The fixture-and-control test we run on real resources, and what it found.
6 min read
I filed my own SR&ED claim. Rebuilding the technological uncertainty story from git history took three weeks. A two-line log would have saved them.
8 min read
A laptop deploy and a CI deploy landed 18 seconds apart. The pipeline went red and production kept running the OLD build. The error never named the consequence.
6 min read
Cost Explorer data can lag up to 24 hours. Inside CloudWise's daily anomaly scan: a real z-score threshold and the test that proves it catches a spike.
5 min read
Our /dashboard cold start took 1,124ms. SnapStart was on and restoring in ~600ms, yet made no measurable change for two deploys. The debugging path.
7 min read
A wrong password on the CloudWise login page returned HTTP 500 on staging and production, unnoticed until an E2E test caught it. Status codes are a contract.
6 min read