You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Heng Chen (JIRA)" <ji...@apache.org> on 2016/02/17 10:39:18 UTC

[jira] [Commented] (HBASE-15280) Scan metrics should not ignore empty rows.

    [ https://issues.apache.org/jira/browse/HBASE-15280?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15150189#comment-15150189 ] 

Heng Chen commented on HBASE-15280:
-----------------------------------

{code}
+    count = scanTable(table, rowa, rowd);
+    metricsRegionServer.getRegionServerWrapper().forceRecompute();
+    assertEquals(count, 2);
+    // It is weired! The deleted rowb not calculated!!
+    metricsHelper.assertCounter("readRequestCount", readRequests + 6, serverSource);
{code}
Table have rows a, b, d
As patch shows,  i  delete row b,  and scan from a to d,  only a, d will return,  but i can't explain why deleted  row b not calculated in readRequestCount now. 
Let me have a deeper insight.

> Scan metrics should not ignore empty rows.
> ------------------------------------------
>
>                 Key: HBASE-15280
>                 URL: https://issues.apache.org/jira/browse/HBASE-15280
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Heng Chen
>         Attachments: HBASE-15280.patch
>
>
> It is come from HBASE-15267.
> we found that as for empty result,  GET increment the readRequestCount, but Scan does not do it.   
> Detail information, please see  HBASE-15267 comments.
> This jira is to fix it.  We should record each request we do.  At least, we could make Scan metrics behavior be consistent with Get. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)