[ https://issues.apache.org/jira/browse/HBASE-19126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16291231#comment-16291231 ]
stack commented on HBASE-19126: ------------------------------- Moving out an important but nice-to-have issue. > [AMv2] RegionStates/RegionStateNode needs cleanup > ------------------------------------------------- > > Key: HBASE-19126 > URL: https://issues.apache.org/jira/browse/HBASE-19126 > Project: HBase > Issue Type: Improvement > Reporter: Yi Liang > Fix For: 2.0.0 > > > // Mutable/Immutable? Changes have to be synchronized or not? > // Data members are volatile which seems to say multi-threaded access is > fine. > // In the below we do check and set but the check state could change before > // we do the set because no synchronization....which seems dodgy. Clear up > // understanding here... how many threads accessing? Do locks make it so one > // thread at a time working on a single Region's RegionStateNode? Lets > presume > // so for now. Odd is that elsewhere in this RegionStates, we synchronize on > // the RegionStateNode instance > Copied from TODO in RegionState.java > Open this jira to track some cleanups for RegionStates/RegionStateNode -- This message was sent by Atlassian JIRA (v6.4.14#64029)