Error: dynamodb system errors metric

SystemErrors is a CloudWatch metric counting DynamoDB or DynamoDB Streams requests that return HTTP 500. It does not mean the metrics service is broken. SDK retries can recover these errors before your application sees a failure, while still adding latency.

1. Inspect AWS/DynamoDB SystemErrors with the TableName and Operation dimension pair and the Sum statistic. A table-only dimension does not match this metric.
2. Correlate the time window with application exceptions, request IDs, retry counts, and client latency. Review FaultInjectionServiceInducedErrors if an AWS FIS experiment is running.
3. Use bounded SDK retries. For an uncertain write outcome, reconcile the item or use an idempotent design before replaying a non-idempotent operation.
4. Investigate sustained failures through AWS Health and AWS Support with Region, table, operation, timestamps, and request IDs.

Do not attempt to reboot DynamoDB or toggle metrics. It is a managed service. Keep throttling and validation failures separate from this metric when diagnosing the incident.

See DynamoDB monitoring and AWS documentation.