[HBASE-25445] Old WALs archive fails in procedure based WAL split

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 3.0.0-alpha-1, 2.4.0, 2.2.6, 2.3.2
• Fix Version/s: 3.0.0-alpha-1, 2.2.7, 2.3.4, 2.5.0, 2.4.1
• Component/s: wal
• Hadoop Flags: Reviewed
• Assignee: Anjan Das
• Reporter: mokai
• Created: 25/Dec/20 07:06
• Updated: 16/Jan/21 10:29
• Resolved: 07/Jan/21 10:34

Description

If 'hbase.wal.dir' and 'hbase.rootdir' are configured to different filesystem, SplitWALRemoteProcedure archived split WAL failed since SplitWALManager using wrong fs instance. SplitWALManager should use WAL corresponding fs instance.

Steps to Reproduce:

• Configure 'hbase.wal.dir' and 'hbase.rootdir' so that they point to different fs instances.
• Start HBase with multiple RS. 
• Create a couple of tables and some rows in them so that the RSs get assigned with some regions. 
• Take any RS with non-zero number of regions offline. 
• Check master logs for "Wrong FS" error as shown in the screenshot attached. 

 

Issue Links

• is caused by: HBASE-21588 Procedure v2 wal splitting implementation
• is cloned by: HBASE-25470 Add unitest for HBASE-25445 - SplitWALRemoteProcedure failed to archive split WAL
• is related to: HBASE-24632 Enable procedure-based log splitting as default in hbase3
• relates to: HBASE-25475 Improve unit test for HBASE-25445 : SplitWALRemoteProcedure failed to archive split WAL
• links to: GitHub Pull Request #2844