[HBASE-22684] The log rolling request maybe canceled immediately in LogRoller due to a race

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 3.0.0-alpha-1, 2.3.0, 2.0.6, 2.2.1, 2.1.6
• Component/s: wal
• Hadoop Flags: Reviewed
• Assignee: Duo Zhang
• Reporter: Duo Zhang
• Created: 12/Jul/19 14:29
• Updated: 07/Aug/19 03:10
• Resolved: 15/Jul/19 14:50

Description

Copy the comments in HBASE-22681.

OK, found another problem in LogRoller, where the rollLog flag is set to false at the end of the run method, right after we set it to true in the logRollRequested method. This cause my test fail after I fixed the unackedAppends problem.

Issue Links

• is blocked by: HBASE-22681 The 'assert highestUnsyncedTxid < entry.getTxid();' in AbstractFWAL.append may fail when using AsyncFSWAL
• relates to: HBASE-22665 RegionServer abort failed when AbstractFSWAL.shutdown hang
• links to: GitHub Pull Request #378

Sub-Tasks

• [branch-1] The log rolling request maybe canceled immediately in LogRoller due to a race: Resolved