search

apache / amoro

Apache Amoro (incubating) is a Lakehouse management system built on open data lake formats.
https://amoro.apache.org/
Apache License 2.0
860 stars 284 forks source link

[Feature]: Provide ability to impersonate to the owner of each table when optimizing #3197

Open byungnam opened 1 month ago

byungnam commented 1 month ago

Description

Currently, Amoro uses the specified UGI - typically, superuser for the optimization tasks. This results in the output files of the optimization task being owned by the superuser instead of the actual owner of the table.

Use case/motivation

No response

Describe the solution

Make an option on catalog which enables impersonation to the owner of each table. If this option is enabled, optimize task will be run with impersonation to the owner of each table.

Subtasks

No response

Related issues

No response

Are you willing to submit a PR?

Code of Conduct

majin1102 commented 1 month ago

The feature seems reasonable.

I wonder whether properties on catalog could solve this situation. Could each table be owned by different owner?

byungnam commented 1 month ago

I wonder whether properties on catalog could solve this situation. Could each table be owned by different owner?

It should be a boolean parameter whether you want to create files with the owner of each table or not. It should not specify any user in catalog property.