AWS batch documentation change
Summary
Clarified retry behavior for SageMaker Training jobs failing due to insufficient capacity, specifying retention policies for failed jobs.
Security assessment
Focuses on reliability and capacity management without security implications. No evidence of vulnerability remediation or security feature addition.
Diff
diff --git a/batch/latest/userguide/service-job-retries.md b/batch/latest/userguide/service-job-retries.md index 2f891acc3..c891ef8e3 100644 --- a//batch/latest/userguide/service-job-retries.md +++ b//batch/latest/userguide/service-job-retries.md @@ -26 +26,3 @@ By configuring appropriate retry strategies, you can improve job success rates a -Service jobs automatically retry certain types of failures, such as insufficient capacity errors, without consuming your configured retry attempts. Your retry strategy primarily handles other types of failures such as algorithm errors or service issues. +When a SageMaker Training job fails due to insufficient capacity, AWS Batch retries the job automatically until capacity becomes available. The most recent SageMaker Training job that failed due to insufficient capacity will be retained, and earlier SageMaker Training jobs that failed due to insufficient capacity will be deleted on a best-effort basis. + +These capacity-based retries don't consume your configured retry attempts. Your retry strategy primarily handles other types of failures such as algorithm errors or service issues that occur after a job enters `STARTING`.