Collect and Verify Container Logs/Metrics Into CloudWatch
Explore methods to collect and verify container logs and metrics in AWS CloudWatch. Understand common causes when telemetry fails to appear, how to check logs and metrics delivery, and troubleshoot issues related to configuration, permissions, agents, and network. Gain confidence in monitoring container workloads and maintaining reliable observability pipelines.
We'll cover the following...
When container logs or metrics don't show up in CloudWatch, the cause is rarely CloudWatch itself. It's almost always a broken hop somewhere in the pipeline between the code and CloudWatch, and every broken hop produces the same symptom: nothing shows up.
Container telemetry splits into two artifacts that move differently. Logs are event records that must be shipped to CloudWatch Logs through a log driver, sidecar, or node agent. Metrics are time series that must be emitted or scraped, then forwarded to CloudWatch metrics, often through an integration layer rather than directly from the container.
These artifacts land in different places, and they're billed on different axes, so enabling everything by default is a real tradeoff. Logs are billed by ingestion volume and storage. Metrics are billed by the number of time series and their resolution.
The diagram below shows how information is shared from workload to CloudWatch.
Verification first
A verification checklist should work whether the workload ...