[HBASE-15271] Spark Bulk Load: Need to write HFiles to tmp location then rename to protect from Spark Executor Failures

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.0
• Fix Version/s: 2.0.0

Release Note

When using the bulk load helper provided by the hbase-spark module, output files will now be written into temporary files and only made available when the executor has successfully completed.

Previously, failed executors would leave their files in place in a way that would be picked up by a bulk load command. This caused retried failures to include spurious copies of some cells.

• Assignee: Theodore michael Malaska
• Reporter: Theodore michael Malaska
• Created: 15/Feb/16 14:43
• Updated: 24/Jun/22 19:33
• Resolved: 09/Mar/16 01:47

Description

With the current code if an executor failure before the HFile is close it will cause problems. This jira will have the files first write out to a file that starts with an underscore. Then when the HFile is complete it will be renamed and the underscore will be removed.

The underscore is important because the load bulk functionality will skip files with an underscore.

Issue Links

• is depended upon by: HBASE-14160 backport hbase-spark module to branch-1 and branch-2
• links to: review board