[ 
https://issues.apache.org/jira/browse/HBASE-13738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14554190#comment-14554190
 ] 

Pankaj Kumar commented on HBASE-13738:
--------------------------------------

INCR old entries are getting removed from memstore when CF version is 1.

But cell upsert having some problem seems, it is retaining two entries always. 
Ideally it should be one.

{ code

  // versions visible to oldest scanner
    int versionsVisible = 0;
    while ( it.hasNext() ) {
      Cell cur = it.next();

      if (cell == cur) {
        // ignore the one just put in
        continue;
      }
      // check that this is the row and column we are interested in, otherwise 
bail
      if (CellUtil.matchingRow(cell, cur) && CellUtil.matchingQualifier(cell, 
cur)) {
        // only remove Puts that concurrent scanners cannot possibly see
        if (cur.getTypeByte() == KeyValue.Type.Put.getCode() &&
            cur.getSequenceId() <= readpoint) {
          if (versionsVisible >= 1) {

 code }

> Scan with RAW type for increment data insertions is displaying only latest 
> two KV's 
> ------------------------------------------------------------------------------------
>
>                 Key: HBASE-13738
>                 URL: https://issues.apache.org/jira/browse/HBASE-13738
>             Project: HBase
>          Issue Type: Bug
>          Components: Scanners
>         Environment: Suse 11 SP3
>            Reporter: neha
>            Assignee: Pankaj Kumar
>            Priority: Minor
>
> [Scenario for Reproducing ]:
> 1. Create an HBase table with single column family by keeping the versions=1 
> (DEFAULT)
> 2. Increment Insertion more than 2 times for the same row and for same 
> qualifier.
> 3. scan the table with raw= true and versions= 10  
> {code}
> scan 'tbl', {RAW => TRUE, VERSIONS => 10}
> {code}
> Expected Result:
> ===============
> Raw scan should result in all the versions until the table flushed



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

Reply via email to