[ https://issues.apache.org/jira/browse/HBASE-23271?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16978252#comment-16978252 ]
qiang Liu commented on HBASE-23271: ----------------------------------- [~psomogyi] could you please review this? thanks in advance. > HFileReader get split point handle empty HFile better > ----------------------------------------------------- > > Key: HBASE-23271 > URL: https://issues.apache.org/jira/browse/HBASE-23271 > Project: HBase > Issue Type: Improvement > Components: regionserver > Affects Versions: 3.0.0, 1.1.7 > Reporter: qiang Liu > Assignee: qiang Liu > Priority: Minor > Labels: easyfix > Attachments: HBASE-23271.patch > > > currenttly if we call org.apache.hadoop.hbase.io.hfile.HFileReaderImpl#midKey > on an empty HFile, we got an exception like > java.io.IOException: HFile empty > since the function return an Optional<Cell> , I think it's better return > Optional.empty() instead of throw An Exception > when a region with muiltiple column family grow big enough to be splited, if > got some empty column family, will got some warn log like this > java.io.IOException: HFile empty at > org.apache.hadoop.hbase.io.hfile.HFileBlockIndex$BlockIndexReader.midkey(HFileBlockIndex.java:334) > sinece exception is catched, split logic will go on and got right result -- This message was sent by Atlassian Jira (v8.3.4#803005)