The Night of the Living Containers: Horror Stories from the DevOps World
summary
Concrete DevOps horror stories about container platforms and costs: an AWS ECS Fargate cluster repeatedly re-pulls images causing 16 TB egress (~$800) for development; a containerized app with a memory leak dies liveness probes and is “fixed” by restarting every 2 hours instead of fixing the bug; hugely oversized Docker images (larger than Windows); a misconfigured CDN that forces asset downloads on first start; and a saturated 1 Gbps site-to-site VPN that drops the application. For each case the talk explains what is broken, which flag or architectural change fixes it, and the visible cost impact.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:21ECS Fargate restarts
- 6:52Memory leaks in containers
- 11:53Dangers of rollback
- 15:45CDN misconfigurations
- 18:14Cloud Native myths
- 23:58Architecture pitfalls
- 25:44Q&A
related talks
49mIn Silicon Valley, Engineers Deploy-and-Tell: Stories About Dockerizing an Application
40mAvoiding Pain When Operating in the Cloud
28mDevOps Road Blocks
31mMigrating a monolith to Cloud-Native (and the stumbling blocks)
26mContainerization crossroads: Experiences-Experiments-Expectations
31mFifteen Years of Excess Has Lead to the DevOps Hangover
56mDevOps in a containerized world
47mContinuous Delivery with Containers: The Good, Bad, & Ugly