HDDS-6709. Fix bucket usedBytes while versioning is true.

[captainzmc]

What changes were proposed in this pull request?

If Bucket versioning is set to true, the key should be able to keep multiple versions. And the keyToDelete in OMKeyCommitRequest will not be deleted. At this time the keyToDelete's usedBytes also should not detracted.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-6709

How was this patch tested?

UT has added.

[captainzmc]

Hi @kuenishi, keyToDelete was added in HDDS-5461. So can you help review this patch? cc @errose28

[errose28]

LGTM, just a minor nit on the test. @captainzmc up to you whether you want to merge this or #3286 first, so you can resolve possible merge conflicts in the one that gets merged second.

[errose28]

Test looks good, but can we use org.junit.jupiter.params.ParameterizedTest for the bucket layout types, instead of new test methods, similar to what the EC tests are doing for replication config in this class?

[captainzmc]

Sure. Let's change test using ParameterizedTest.

[kerneltime]

Nit: Might be cleaner to combine the identical if conditions here and line 158

[captainzmc]

Good point. Let's combine the identical.

[kerneltime]

Might as well combine all the versioned bucker handling here and line 224

[kuenishi]

Good catch and thank you for fixing it! I'm afraid the same bug resides in committing multipart objects via S3 API: https://github.com/apache/ozone/blob/d1117ebe1689eb51695594c4a37a6f1a1672523f/hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/request/s3/multipart/S3MultipartUploadCompleteRequest.java#L225-L244

Fixing it in this issue or another one, either would be fine for me.

[captainzmc]

Fixing it in this issue or another one, either would be fine for me.

Thanks @kuenishi for the tip, let's fix it in this PR.

[kuenishi]

Thank you, all :)

[captainzmc]

Thanks @errose28 @kerneltime @kuenishi for the review. Let's merge this.