[Bug]: If the optimizing of a Mixed Hive table fails to alter the Hive location, the files will be moved to the old Hive location.

[wangtaohz]

What happened?

When optimizing the Mixed Hive Format table but failing to alter the hive location, the following full-optimizing will move files to the old Hive location. This brings two problems

• If the asynchronous thread successfully alters the Hive location, users will see a decrease in the data in Hive, similar to phantom reads.
• The commit of the following full-optimizing will fail, because the old hive location contains a mixture of files from new and old snapshots. The exception will be like

Affects Versions

0.5.x,0.4.x,0.3.x

What engines are you seeing the problem on?

Core

How to reproduce

This issue occurs only when failing to alter the Hive location.

To reproduce this scenario

1. We can deliberately change the hive location back to its original location after the full-optimizing committing.
2. Then insert some data, and wait for the next full-optimizing move files to the old hive location.
3. Then trigger another full-optimizing, it will fail when committing.

Relevant log output

No response

Anything else

No response

Are you willing to submit a PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[wangtaohz]

To fix this issue, we should make sure that the Hive location is correct before committing of optimizing.