[HUDI-5220] failed to snapshot query in hive when query a empty partition

• Type: Bug
• Status: In Progress
• Priority: Major
• Resolution: Unresolved
• Component/s: hive
• Labels: pull-request-available
• Assignee: Unassigned
• Reporter: yuehanwang
• Created: 15/Nov/22 06:43
• Updated: 18/Oct/23 23:29

Description

When query a empty partition hive will return a empty file in split path. This path will be added as a NonHoodieInputPaths. In this case HoodieParquetRealtimeInputFormat read a file split rather than a RealtimeSplit. Throw a exception:

HoodieRealtimeRecordReader can only work on RealtimeSplit and not with hdfs://test-cluster/tmp/hive/20220520/hive/4273589d-49be-4a60-9890-a29660d81927/hive_2022-11-14_11-32-41_221_5694963332005566615-17/-mr-10004/74adf5bb-b07e-4eac-a90b-1b5a7fc3d5c4/emptyFile:0+466

Issue Links

• links to: GitHub Pull Request #7206