[ https://issues.apache.org/jira/browse/HBASE-20658?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16508784#comment-16508784 ]
Josh Elser commented on HBASE-20658: ------------------------------------ Yeah, I think that's what Duo mean, [~an...@apache.org]. Truncate was a synchronous operation, so, if we released the latch early, I think a 1.x client would return too fast. However, modify table was an async operation so that reason doesn't hold for it. However, as you see, we added the synchronous lock for HBASE-19953. I did leave a [comment|https://issues.apache.org/jira/browse/HBASE-19953?focusedCommentId=16370671&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16370671] which might be relevant. > Release latch early for ModifyTable and TruncateTable procedure. > ---------------------------------------------------------------- > > Key: HBASE-20658 > URL: https://issues.apache.org/jira/browse/HBASE-20658 > Project: HBase > Issue Type: Bug > Reporter: Ankit Singhal > Assignee: Ankit Singhal > Priority: Major > > It seems we are releasing latch in Modify and Truncate table procedures > during completion cleanup which make the whole procedure to execute > synchronously. -- This message was sent by Atlassian JIRA (v7.6.3#76005)