Presently key path normalization and prefix(
HDDS-2939) feature configurations ozone.om.enable.filesystem.paths and ozone.om.metadata.layout are at OM cluster-level. Using these configurations, OM cluster can be configured and started to satisfy only a specific use case, either ‘Pure S3' or ‘FileSystem’ semantics.
This is an umbrella jira talking about moving the above two cluster-level configurations to a bucket level configuration so that single OM will efficiently hold multiple buckets with different modes at at time. A user who wants to satisfy ‘Pure S3' use case, he can define the bucket in that mode and other user who wants to satisfy ‘FileSystem’ use case, he can define the bucket in that mode.
Please see/create sub-tasks for the overall changes for realising this efforts. Thanks!