[ https://issues.apache.org/jira/browse/TRAFODION-2839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16291603#comment-16291603 ]
ASF GitHub Bot commented on TRAFODION-2839: ------------------------------------------- GitHub user DaveBirdsall opened a pull request: https://github.com/apache/incubator-trafodion/pull/1345 [TRAFODION-2839] Change Incr UPDATE STATS to not use CBFs by default Change the default for CQD USTAT_INCREMENTAL_UPDATE_STATISTICS from 'ON' to 'SAMPLE'. This causes the incremental update statistics logic to skip using Counting Bloom Filters, but instead always rebuild the histograms from the incremental sample. By doing so, we avoid accumulating CBF files in the $HOME/cbfs directories. As a side benefit, it turns out there is an elapsed time improvement to incremental update statistics in doing this. (Making this change was suggested by @blfritch a year ago actually, for performance reasons.) You can merge this pull request into a Git repository by running: $ git pull https://github.com/DaveBirdsall/incubator-trafodion Trafodion2839 Alternatively you can review and apply these changes as the patch at: https://github.com/apache/incubator-trafodion/pull/1345.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #1345 ---- commit 149d1cfed176c511f8ac4826d840af7b5894c625 Author: Dave Birdsall <dbirds...@apache.org> Date: 2017-12-14T21:05:43Z [TRAFODION-2839] Change Incr UPDATE STATS to not use CBFs by default ---- > compGeneral/TEST023 leaves a bunch of stuff in /home/xxx/cbfs directory > ----------------------------------------------------------------------- > > Key: TRAFODION-2839 > URL: https://issues.apache.org/jira/browse/TRAFODION-2839 > Project: Apache Trafodion > Issue Type: Bug > Components: sql-cmp > Environment: All > Reporter: David Wayne Birdsall > Assignee: David Wayne Birdsall > > The UPDATE STATISTICS incremental code leaves CBF files in the /home/xxx/cbfs > directory. These stick around even after tables are deleted. There should be > a way to clean these up. -- This message was sent by Atlassian JIRA (v6.4.14#64029)