This documentation references Amazon ParallelCluster. For running HPC and AI workloads, we recommend Amazon Parallel Computing Service (Amazon PCS)
Working with Amazon S3
You can configure Amazon ParallelCluster's access to Amazon S3 through the HeadNode / Iam / S3Access and Scheduling / SlurmQueues / - Name / Iam / S3Access parameters in the Amazon ParallelCluster configuration.
Examples
The following example configures read-only access to all objects in firstbucket/read_only/ and
read/write access to all objects in .secondbucket/read_and_write/
... HeadNode: ... Iam: S3Access: - BucketName:firstbucketKeyName:read_only/*EnableWriteAccess:false- BucketName:secondbucketKeyName:read_and_write/*EnableWriteAccess:true...
The next example configures read-only access to all objects in folder in any bucket (*) in the account.read_only/
... HeadNode: ... Iam: S3Access: - BucketName:*KeyName:read_only/*EnableWriteAccess:false...
The final example configures read_only access to all buckets and objects in the account.
... HeadNode: ... Iam: S3Access: - BucketName: * ...