[ https://issues.apache.org/jira/browse/HBASE-3601?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack resolved HBASE-3601. -------------------------- Resolution: Fixed Assignee: stack Committed to TRUNK (though this an issue in 0.90.x too... when new master in mix). > TestMasterFailover broken in TRUNK > ---------------------------------- > > Key: HBASE-3601 > URL: https://issues.apache.org/jira/browse/HBASE-3601 > Project: HBase > Issue Type: Bug > Reporter: stack > Assignee: stack > > After HBASE-3573, went in, TestMasterFailover broke. The change in shutdown > technique revealed an issue with our in-memory accounting when a master joins > an already cluster; we don't add .META. and -ROOT- to our set of online > regions in the new master so could make for some interesting issues as the > new master progressed (Previous shutdown did a count of remaining servers, > new shutdown process looks at in-memory state to see if only catalog carrying > regionservers online... this is what was going out of whack in new master). -- This message is automatically generated by JIRA. - For more information on JIRA, see: http://www.atlassian.com/software/jira