HDDS-10367. Fix possible NullPointerException in listKeysLight method

[ivanzlenko]

What changes were proposed in this pull request?

Fix possible NullPointerException which can occur in some cases while invoking listKeysLight from OzoneManagerProtocolClientSideTranslatorPB.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-10367

How was this patch tested?

This patch is test manually.
Separate epic created to cover testing of hdfs client with S3A protocol: https://issues.apache.org/jira/browse/HDDS-10381

[adoroszlai]

Thanks @ivanzlenko for the patch.

Please add test case (for non-light listStatus) in TestListStatus#testSortedListStatus.

[ivanzlenko]

@adoroszlai I've added additional test cases but looking at test coverage all changes were covered already. Looks like some of the issues with s3a protocol were fixed in HDDS-9762, but my fixes should prevent future cases where we accidentally could pass 'null' for startKey.

[adoroszlai]

Thanks @ivanzlenko for updating the patch.

[kerneltime]

cc @tanvipenumudy

[myskov]

lgtm

[adoroszlai]

Thanks @ivanzlenko for restoring package in the test class.

Interestingly, the new test case passes without the added null checks, so it does not seem to exercise the code path being changed.

[ivanzlenko]

@adoroszlai we have some preliminary checks for nulls in intermediary code. So this test cases here for good measure. I believe we would've caught HDDS-9762 earlier having cases with null value.

[adoroszlai]

Since this is also a ListStatusRequest, I think we need to set "" in else branch here, too.

[sumitagrawl]

@ivanzlenko Thanks for working over this, its fix for consistency with listKeys, but do not have any NullPointerException as startKey as null is expected value.
And for improving test cases.

[adoroszlai]

Thanks @ivanzlenko for the patch, @myskov, @sumitagrawl for the review.

[ivanzlenko]

@myskov@sumitagrawl@adoroszlai thanks for review!