[ https://issues.apache.org/jira/browse/HBASE-3573?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13000622#comment-13000622 ]
stack commented on HBASE-3573: ------------------------------ Yeah. The notion is that we set the cluster shutdown flag then wait. When regionservers notice the cluster shutdown flag, they will close their user regions. Regionservers shut themselves down if all regions are closed and the cluster shutdown flag is set. So, a regionserver carrying user regions only will just go shut itself down. Servers carrying catalog regions will stick around. So, the cited check is just waiting till the only servers running are those carrying catalog regions. When we are in this state, the thinking is that near all user regions are closed.... so its safe to close out the catalog regions. So, yeah, they could be holding user regions but should be ok... the user regions are at least closed if not closed. This is basically how shutdown currently works. Trying to do minimal change for now. Can come back and review shutdown process after the heartbeat removal (It might make sense having the close go via zk instead of skipping it as we currently do -- using bulk close facility) > Move shutdown messaging OFF hearbeat; prereq for fix of hbase-1502 > ------------------------------------------------------------------ > > Key: HBASE-3573 > URL: https://issues.apache.org/jira/browse/HBASE-3573 > Project: HBase > Issue Type: Task > Reporter: stack > Assignee: stack > Fix For: 0.92.0 > > Attachments: 3573.txt, 3573.txt > > -- This message is automatically generated by JIRA. - For more information on JIRA, see: http://www.atlassian.com/software/jira