Architecture note2 lab guides2 lab guides
CloudTrail · AWS CloudTrail
API audit logs, trails vs Event history, management and data events, and the StopLogging failure that goes unnoticed until the investigation.
Outcome
- Created an S3 log bucket with a CloudTrail-only write policy
- Created a multi-region trail with log file validation
- Started logging and read
GetTrailStatus - Looked up
CreateBucketin Event history before S3 delivery - Confirmed log objects under
AWSLogs/.../CloudTrail/ - Added a scoped S3 data-event selector (not
*) - Generated
PutObjectdata-plane activity - Attached CloudWatch Logs delivery with a dedicated IAM role
- Created a metric filter and alarm for
StopLogging/DeleteTrail - Stopped logging and showed Event history still records the call
- Broke the bucket policy and observed
LatestDeliveryError - Listed digest prefixes and ran
validate-logs - Cleaned up the trail, log group, role, and buckets