[ https://issues.apache.org/jira/browse/HBASE-4476?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13117749#comment-13117749 ]
jirapos...@reviews.apache.org commented on HBASE-4476: ------------------------------------------------------ ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/2050/ ----------------------------------------------------------- (Updated 2011-09-29 23:45:46.477105) Review request for hbase and Kannan Muthukkaruppan. Changes ------- Linking to HBASE-4476. Summary ------- We found this in ScanWildcardColumnTracker: // new col < oldcol // if (cmp < 0) { // WARNING: This means that very likely an edit for some other family // was incorrectly stored into the store for this one. Continue, but // complain. LOG.error("ScanWildcardColumnTracker.checkColumn ran " + "into a column actually smaller than the previous column: " + This went under the radar in our dark launch cluster when a column family name was misspelled first, but then was "renamed" by renaming directories in the HBase storage directory tree. We ended up with inconsistent data, but compactions still succeeded most of the time, likely discarding part of input data. This addresses bug HBASE-4476. https://issues.apache.org/jira/browse/HBASE-4476 Diffs ----- http://svn.apache.org/repos/asf/hbase/trunk/src/main/java/org/apache/hadoop/hbase/regionserver/ColumnTracker.java 1176613 http://svn.apache.org/repos/asf/hbase/trunk/src/main/java/org/apache/hadoop/hbase/regionserver/ScanQueryMatcher.java 1176613 http://svn.apache.org/repos/asf/hbase/trunk/src/main/java/org/apache/hadoop/hbase/regionserver/ScanWildcardColumnTracker.java 1176613 http://svn.apache.org/repos/asf/hbase/trunk/src/main/java/org/apache/hadoop/hbase/regionserver/Store.java 1176613 http://svn.apache.org/repos/asf/hbase/trunk/src/main/java/org/apache/hadoop/hbase/regionserver/StoreScanner.java 1176613 http://svn.apache.org/repos/asf/hbase/trunk/src/test/java/org/apache/hadoop/hbase/regionserver/TestExplicitColumnTracker.java 1176613 http://svn.apache.org/repos/asf/hbase/trunk/src/test/java/org/apache/hadoop/hbase/regionserver/TestScanWildcardColumnTracker.java 1176613 Diff: https://reviews.apache.org/r/2050/diff Testing ------- Run unit tests. Will also push to a 5-node dev cluster. Thanks, Mikhail > Compactions must fail if column tracker gets columns out of order > ----------------------------------------------------------------- > > Key: HBASE-4476 > URL: https://issues.apache.org/jira/browse/HBASE-4476 > Project: HBase > Issue Type: Bug > Reporter: Mikhail Bautin > Assignee: Mikhail Bautin > Fix For: 0.92.0, 0.94.0 > > > We found this in ScanWildcardColumnTracker: > // new col < oldcol > // if (cmp < 0) { > // WARNING: This means that very likely an edit for some other family > // was incorrectly stored into the store for this one. Continue, but > // complain. > LOG.error("ScanWildcardColumnTracker.checkColumn ran " + > "into a column actually smaller than the previous column: " + > This went under the radar in our dark launch cluster when a column family > name was misspelled first, but then was "renamed" by renaming directories in > the HBase storage directory tree. We ended up with inconsistent data, but > compactions still succeeded most of the time, likely discarding part of input > data. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira