AWS ECS monitoring
ECS tasks, explained.
A purpose-built sidecar polls the ECS metadata v4 endpoint — so Tracegrid sees task-level OOM and failures on both Fargate and EC2 launch types.
What Tracegrid catches in ECS
Task OOM
An ECS task exceeds its memory and is stopped.
The sidecar reads the kernel oom_kill counter from the task — OOM is detected even on Fargate where /proc access is limited.
CloudWatch log anomalies
Error patterns surface in the task's log group.
Tracegrid matches known-bad patterns and explains them without exporting your full logs.
Scaling events
Service desired-count changes or tasks churn unexpectedly.
Tracegrid correlates scaling activity with failures so flapping is obvious.
Health-check failures
ALB/target-group health checks fail and tasks are replaced.
Tracegrid reports which check failed and the replacement pattern.
How it works
Add the Tracegrid sidecar container to your task definition. It polls ECS Metadata v4 and the kernel OOM counter, reporting incidents the same way the Linux and Kubernetes agents do — one dashboard for everything.
Start monitoring ECS free
Start free monitoring1 host free forever. 15-day full trial on paid plans. No credit card.