You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@phoenix.apache.org by "ramkrishna.s.vasudevan (JIRA)" <ji...@apache.org> on 2015/01/15 08:04:34 UTC
[jira] [Commented] (PHOENIX-1584) Ensure stats are correct if
simultaneous split and merge
[ https://issues.apache.org/jira/browse/PHOENIX-1584?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14278360#comment-14278360 ]
ramkrishna.s.vasudevan commented on PHOENIX-1584:
-------------------------------------------------
When the stats is calculated during compaction it is an absolute measure. But during splits we have an approximation while calculating the percentage. That leads to this variation.
> Ensure stats are correct if simultaneous split and merge
> --------------------------------------------------------
>
> Key: PHOENIX-1584
> URL: https://issues.apache.org/jira/browse/PHOENIX-1584
> Project: Phoenix
> Issue Type: Sub-task
> Reporter: James Taylor
> Assignee: ramkrishna.s.vasudevan
>
> During splits, we update the rowCount and byteCount via StatisticsWriter.splitStats() and at the same time compaction also updates the stats. So we are not able to get a consistent result for the rowCount and byteCount under these conditions. So for now these are not asserted when doing a select from the STATS table in the statistics unit tests. However on the client side it is asserted. All tests passes with this.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)