[ https://issues.apache.org/jira/browse/HBASE-21017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16592572#comment-16592572 ]
Duo Zhang commented on HBASE-21017: ----------------------------------- I think this may happen even if we do not allow assigning an already onlined region. It depends on the implementation of RSProcedureDispatcher. If we may fail a OpenRegionProcedure even if the request has been sent to the RS, this race could happen. I think it is no harm to make it more robust. Just lock the RegionStateNode when executing OpenRegionProcedure. > Revisit the expected states for open/close > ------------------------------------------ > > Key: HBASE-21017 > URL: https://issues.apache.org/jira/browse/HBASE-21017 > Project: HBase > Issue Type: Sub-task > Components: amv2 > Reporter: Duo Zhang > Assignee: Duo Zhang > Priority: Major > Fix For: 3.0.0, 2.2.0 > > Attachments: HBASE-21017-debug.patch, HBASE-21017.patch > > -- This message was sent by Atlassian JIRA (v7.6.3#76005)