Hard links and exporting to Amazon S3 - FSx for Lustre
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Hard links and exporting to Amazon S3

If automatic export (with NEW and CHANGED policies) is enabled on a DRA in your file system, each hard link contained within the DRA is exported to Amazon S3 as a separate S3 object for each hard link. If a file with multiple hard links is modified on the file system, all of the copies in S3 are updated, regardless of which hard link was used when changing the file.

If hard links are exported to S3 using data repository tasks (DRTs), each hard link contained within the paths specified for the DRT is exported to S3 as a separate S3 object for each hard link. If a file with multiple hard links is modified on the file system, each copy in S3 is updated at the time the respective hard link is exported, regardless of which hard link was used when changing the file.

Important

When a new FSx for Lustre file system is linked to an S3 bucket to which hard links were previously exported by another FSx for Lustre file system, Amazon DataSync, or Amazon FSx File Gateway, the hard links are subsequently imported as separate files on the new file system.

A released file is a file whose metadata is present in the file system, but whose content is only stored in S3. For more information on released files, see Releasing files.

Important

The use of hard links in a file system that has data repository associations (DRAs) is subject to the following limitations:

  • Deleting and recreating a released file that has multiple hard links may cause the content of all hard links to be overwritten.

  • Deleting a released file will delete content from all hard links that reside outside of a data repository association.

  • Creating a hard link to a released file whose corresponding S3 object is in either of the S3 Glacier Flexible Retrieval or S3 Glacier Deep Archive storage classes will not create a new object in S3 for the hard link.