HDDS-4232. Use single thread for KeyDeletingService.

[lokeshj1703]

What changes were proposed in this pull request?

KeyDeletingService scan the keys from a particular rocksdb table and sends deletion request to SCM. Every thread would scan the table and send deletion requests. This can lead to multiple deletion request for a particular block. There is currently no way to distribute the keys to be deleted amongst multiple threads.

What is the link to the Apache JIRA

HDDS-4232

How was this patch tested?

This lead to failure in TestBlockDeletion. Please refer 49b610a

[amaliujia]

why not just change 2 to 1?

[lokeshj1703]

Addressed it in latest commit.

[amaliujia]

Thanks!

[bharatviswa504]

Hi @lokeshj1703
HDDS-4120 This Jira changes the behavior of openKeyCleanup service, and does not call SCM directly, it moves the keys to deleteTable by making a write request to OM. And later it will be picked by KeyDeletingService and it will be only the one to send delete keys request to SCM from OM.

Do you think, we need this change still even after this also for openKeyCleanup?

[lokeshj1703]

/pending

[github-actions]

Marking this issue as un-mergeable as requested.

Please use /ready comment when it's resolved.

Please note that the PR will be closed after 21 days of inactivity from now. (But can be re-opened anytime later...)

[lokeshj1703]

The PR requires change in 1414 to address the test failure.

[lokeshj1703]

HDDS-4120 This Jira changes the behavior of openKeyCleanup service, and does not call SCM directly, it moves the keys to deleteTable by making a write request to OM. And later it will be picked by KeyDeletingService and it will be only the one to send delete keys request to SCM from OM.

@bharatviswa504 Thanks for providing the info! In that case we do not need the changes for openKeyCleanupService. I have addressed it in latest commit.

[lokeshj1703]

/ready

Blocking review request is removed.

[adoroszlai]

Thanks @lokeshj1703 for the fix and @amaliujia and @bharatviswa504 for the review.