[HBASE-4552] multi-CF bulk load is not atomic across column families

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.92.0
• Fix Version/s: 0.92.0
• Component/s: regionserver
• Hadoop Flags: Incompatible change

Release Note

HRegion.bulkLoadHFile() has been removed in 0.92 and TRUNK.
HRegion.bulkLoadHFiles() is introduced for multi-family bulk load.

• Assignee: Jonathan Hsieh
• Reporter: Todd Lipcon
• Created: 07/Oct/11 05:21
• Updated: 16/Nov/16 21:02
• Resolved: 01/Nov/11 17:06

Description

Currently the bulk load API simply imports one HFile at a time. With multi-column-family support, this is inappropriate, since different CFs show up separately. Instead, the IPC endpoint should take a of CF -> HFiles, so we can online them all under a single region-wide lock.

Issue Links

• is cloned by: HBASE-4718 Backport HBASE-4552 to 0.90 branch
• is depended upon by: HBASE-4652 Mechanism for atomic recovery from partial failure when atomically bulk-loading multi-CF rows.
• is related to: HBASE-3782 Multi-Family support for bulk upload tools causes File Not Found Exception

Sub-Tasks

• Add atomic bulk load function to region server: Closed
• Update LoadIncrementalHFiles to use atomic bulk load RS mechanism: Closed
• Remove old single bulkLoadHFile method: Closed