You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@gobblin.apache.org by "William Lo (Jira)" <ji...@apache.org> on 2024/06/26 00:18:00 UTC

[jira] [Created] (GOBBLIN-2101) Manifest Distcp pre-create directories do not have the correct owner/group in recursive folder creation

William Lo created GOBBLIN-2101:
-----------------------------------

             Summary: Manifest Distcp pre-create directories do not have the correct owner/group in recursive folder creation
                 Key: GOBBLIN-2101
                 URL: https://issues.apache.org/jira/browse/GOBBLIN-2101
             Project: Apache Gobblin
          Issue Type: Bug
            Reporter: William Lo


Owners and groups are not properly propagated in Manifest distcp where the owner and group of a nested child directory will only be applied to the child dir, and any directory created by fs.mkdirs() will have the proxy user (running user) owner/group. We want to keep this as consistent as possible as distcp.

The suggested approach is to treat this the same as the writer code where folders and ancestors are created with the correct permission tree. We want to create these folders in the data.publisher dir and then rename the files onto them, which will avoid both the concurrency bug where multiple files are renamed at once and also follow up with a SetPermission Commit step (which was removed incorrectly previously) in order to reset the execute bit for folders.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)