Real-world ECS task placement failure scenarios, their causes, and how to fix them.
ECS gives you fine control over where tasks run — but misconfigurations, capacity gaps, and subtle AWS behaviors can keep tasks stuck in PENDING.
This repo collects real scenarios from the field, distilled into quick-reference pages.
| Scenario | Symptom | Quick Fix |
|---|---|---|
| Insufficient Resources | Tasks stuck pending, insufficient CPU/memory |
Add capacity or reduce reservations |
| Constraint Mismatch | First task runs, others pending | Align strategy with constraints |
| Port Conflicts | RESOURCE:PORTS error |
Use dynamic port mapping |
| Subnet IP Exhaustion | Fargate tasks fail to start | Add subnets / expand CIDR |
| Empty Capacity Provider | EMPTY CAPACITY PROVIDER |
Start instances or fix provider |
| Missing Attributes | Missing capability errors | Update AMI / match OS & features |
We welcome new scenarios, diagrams, and fixes. See CONTRIBUTING.md for details.