hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Lars Hofhansl (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HBASE-8759) Family Delete Markers not getting purged after major compaction
Date Thu, 20 Jun 2013 09:05:20 GMT

    [ https://issues.apache.org/jira/browse/HBASE-8759?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13688988#comment-13688988
] 

Lars Hofhansl edited comment on HBASE-8759 at 6/20/13 9:05 AM:
---------------------------------------------------------------

NP. I should also be a bit more explicit how family delete markets are actually deleted. The
logic is this:
Each compaction registers the timestamp of the oldest put in the created hfile.
The next (major) compaction then removes all family markers that are older than the oldest
put. So as long as the client does not keep backdating puts, eventually all family markers
drop out of the compacted files.
                
      was (Author: lhofhansl):
    NP. I should also be a bit more explicit how family delete markets are actually deleted.
The logic is this:
Each compaction registers the timestamp of the oldest put in the created hfile.
The next (major) compaction then removes all family markers that are older than the oldest
put. So as long as the client does not keep puts, eventually all family markers drop out of
the compacted files.
                  
> Family Delete Markers not getting purged after major compaction
> ---------------------------------------------------------------
>
>                 Key: HBASE-8759
>                 URL: https://issues.apache.org/jira/browse/HBASE-8759
>             Project: HBase
>          Issue Type: Bug
>          Components: Compaction
>    Affects Versions: 0.94.7
>            Reporter: Mujtaba Chohan
>            Priority: Minor
>
> On table with VERSIONS => '1', KEEP_DELETED_CELLS => 'true'. Family Delete Markers
does not get purged after put > delete > major compaction (they keep on incrementing
after every put > delete > major compaction)
> Following is the raw scan output after 10 iterations of put > delete > major compaction.
> ROW                                       COLUMN+CELL                               
                                                                             
> A                                        column=CF:, timestamp=1371512706683, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512706394, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512706054, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512705763, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512705457, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512705149, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512704836, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512704518, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512704162, type=DeleteFamily
                                                                 
> A                                        column=CF:, timestamp=1371512703779, type=DeleteFamily
                                                                 
> A                                        column=CF:COL, timestamp=1371512706682, value=X

> [~lhofhansl]
> Code to repro this issue:
> http://phoenix-bin.github.io/client/code/delete.java

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message