errose28 commented on PR #3235: URL: https://github.com/apache/ozone/pull/3235#issuecomment-1089581072
Thanks for the quick reply. > Default key name in the openKeyTable /volume/bucket/key/clientID. FSO key name in the openFileTable parentID/key/clientID. This is correct, but the service is cleaning out the table regardless of the key format. It iterates the correct table, and posts deletions for the keys it finds. With the current implementation, we will find a key, split it into parts, then re-assemble it again later to get the key we need to delete. As long as we know what table we got the key from, we can leave it assembled and still retrieve it again. > That's why I also included BucketLayout in OmOpenKeyInfo. Yes, I agree we need this field in the OmOpenKeyInfo, but it is currently set using `OmMetadataManagerImpl#getBucketLayout` which is a stub always returning DEFAULT. Even if the method were corrected to take a bucket name and check the layout, this information could be incorrect in the scenarios given in my previous comment. IMO there should be two methods, maybe with a helper method to handle common code: `OmMetadataManagerImpl#getExpiredOpenKeys` and `OmMetadataManagerImpl#getExpiredOpenFiles`. Each method scans only its corresponding table, and sets the `bucketLayout` flag according to the table it scanned. This also seems consistent with the plan described in HDDS-6491, where the cleanup service must be able to choose which table should be scanned on the current run. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
