You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@phoenix.apache.org by "Alex Batyrshin (Jira)" <ji...@apache.org> on 2019/12/11 11:34:00 UTC

[jira] [Created] (PHOENIX-5611) Inaccurate MutationState numRows and estimatedSize counting

Alex Batyrshin created PHOENIX-5611:
---------------------------------------

             Summary: Inaccurate MutationState numRows and estimatedSize counting
                 Key: PHOENIX-5611
                 URL: https://issues.apache.org/jira/browse/PHOENIX-5611
             Project: Phoenix
          Issue Type: Bug
    Affects Versions: 4.14.2
            Reporter: Alex Batyrshin
         Attachments: numRows.patch

Hello,
I found that numRows decreased by HBase mutations count instead of SQL rows count. In case of NULL in UPSERT there will be 1 SQL row  == 2 HBase mutation.

UPSERT INTO test VALUES ('foo', NULL) => HBase Put Mutation + HBase Delete Mutation

Also I found that estimatedSize could be calculated via KeyValueUtil.getEstimatedRowMutationSize and via RowMutationState.calculateEstimatedSize. Last one doesn't take into account length of row keys, so result KeyValueUtil.getEstimatedRowMutationSize != RowMutationState.calculateEstimatedSize



--
This message was sent by Atlassian Jira
(v8.3.4#803005)