🩺
CloudWatch & X-Ray
Troubleshooting and OptimizationObservability: metrics, logs, alarms, and distributed tracing.
CloudWatch Metrics & Alarms
- ▸PutMetricData publishes custom application/business metrics.
- ▸Alarms watch a metric/expression and change state on threshold breach, triggering SNS/Auto Scaling actions.
- ▸Standard resolution: 1 minute; high resolution: down to 1 second.
CloudWatch Logs
- ▸Log groups/streams collect application and Lambda logs automatically.
- ▸Logs Insights: interactive query language to search/filter/aggregate logs directly.
- ▸Subscription filters can stream logs in real time to Lambda, Kinesis, or OpenSearch.
X-Ray
- ▸Traces requests across service boundaries; builds a visual service map.
- ▸Segments (per service) and subsegments (per call/operation) capture timing and errors.
- ▸Sampling rules control what percentage of requests are traced, to control cost/volume.
Common failure signatures
- ▸Lambda "Task timed out": raise the timeout and/or optimize slow downstream calls.
- ▸DynamoDB throttling with skewed partition metrics: hot partition — fix key design or use on-demand.
- ▸API Gateway 502: malformed Lambda proxy response. 504: backend took too long.
- ▸SQS poison-pill messages: configure a DLQ with maxReceiveCount to isolate them.