Ideas for
your next move.
Practical thinking for teams building software, securing systems and operating modern infrastructure. Explore our engineering field notes.
500 insights · Page 20 of 42
ECR Lifecycle Policies: Preview the Release Impact
Review ECR lifecycle rules against deployed digests, rollback requirements, tag patterns, and multi-architecture manifests before applying retention.
Read field note ↗Cognito M2M Tokens: Connect Custom Scopes to API Actions
Set up Cognito client-credentials authorization with a confidential client, explicit resource-server scopes, secure token handling, and API enforcement tests.
Read field note ↗Cognito Callback Errors: Trace the Exact Redirect URI
Resolve Cognito callback failures by comparing app-client registration, authorization and token requests, proxy-derived URLs, and a complete test sign-in.
Read field note ↗EventBridge Scheduler: Make Time-Zone Behavior Explicit
Set EventBridge Scheduler recurrence with an explicit time zone, daylight-saving policy, flexible window, and evidence that scheduled work completed.
Read field note ↗SQS FIFO: Find the Message Group That Limits Progress
Diagnose FIFO queue backlog through message-group distribution, in-flight work, deduplication identifiers, and the business ordering boundary.
Read field note ↗SQS Visibility: Match the Lease to the Work
Reduce avoidable SQS duplicate work by measuring processing time, extending visibility deliberately, and verifying deletion with the current receipt handle.
Read field note ↗SNS Filters: Match the Scope to the Published Message
Diagnose SNS messages missing from a subscription by comparing filter scope, attribute types, JSON body shape, and controlled matching and nonmatching events.
Read field note ↗Step Functions Retry and Catch: Preserve Business Effects
Design Step Functions error handling around named failures, bounded retries, retained input, and verification of business effects after ambiguous timeouts.
Read field note ↗Lambda Python Layers: Match Runtime and Architecture
Diagnose Lambda layer import failures by checking archive paths, Python versions, Linux compatibility, native architecture, and the exact published layer version.
Read field note ↗Lambda Async Destinations: Verify the Failure Record
Trace Lambda asynchronous failures from retry settings to destination permissions, record delivery, and an idempotent business recovery procedure.
Read field note ↗Lambda Concurrency: Separate Capacity from Warm Environments
Choose Lambda reserved and provisioned concurrency using measured demand, downstream limits, alias routing, and evidence of throttling or provisioned spillover.
Read field note ↗DynamoDB Streams: Recover a Lambda Batch Deliberately
Investigate DynamoDB stream backlog with iterator age, record-level failures, partial batch configuration, and a durable recovery plan for discarded work.
Read field note ↗