yuqi1129 commented on issue #13695: URL: https://github.com/apache/gravitino/issues/13695#issuecomment-6137394178
@pithecuse527 Thanks, you are right. This appears to be a legacy design. The type tokens (`METALAKE`, `CATALOG`, etc.) were added to distinguish resource types in cache keys. For example, the name path alone cannot tell whether an entity under a schema is a fileset, table, topic, model, or another resource type. For the metalake itself, however, the metalake name is sufficient. We can remove the `METALAKE` token and use `<ml>␟` as its key, keeping the trailing separator so prefix invalidation respects the metalake name boundary. This restores the parent-key-as-prefix behavior without needing a separate invalidation-prefix helper. Please also cover both invalidation paths: the local metalake drop hook needs to notify the authorizer of the name-to-ID mapping change, and peer nodes need to invalidate the entire metalake scope when replaying the `METALAKE DROP` event. Removing the `METALAKE` token alone does not complete this issue. Please add tests for both paths, including drop and recreation with same-named children. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
