[ 
https://issues.apache.org/jira/browse/PHOENIX-2696?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Ankit Singhal updated PHOENIX-2696:
-----------------------------------
    Attachment: PHOENIX-2696.patch

I think the deletes will still be in WAL and data on disk when major_compaction 
was run. so that's why they are not visible.
[~mujtabachohan], can you try flushing your deletes as well before starting 
major_compaction. then it should not record any guideposts and delete the old 
one as well.

[~jamestaylor], should we handle such case of deleting stale stats during 
update statistics command also (like attached patch) when the region becomes 
shorter due to deletes and there is no stats collected ?


> Major compaction generating wrong guideposts
> --------------------------------------------
>
>                 Key: PHOENIX-2696
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-2696
>             Project: Phoenix
>          Issue Type: Bug
>            Reporter: Mujtaba Chohan
>            Priority: Minor
>         Attachments: PHOENIX-2696.patch
>
>
> With empty stats table. _Update Statistics_ correctly generate no guideposts 
> in SYSTEM.STATS table for a table containing 0 rows (this table is major 
> compacted after 200K rows upserted/deleted)
> Again with empty stats table, after _major compaction_ of this table, 
> following stats are generated. These stats are generated even after multiple 
> major compactions (guidepost width is set to 5M)
> {code}
> +---------------------+----------------+-----------------+--------------------+--------------------------+------------------------+
> |    PHYSICAL_NAME    | COLUMN_FAMILY  | GUIDE_POST_KEY  | GUIDE_POSTS_WIDTH  
> |  LAST_STATS_UPDATE_TIME  | GUIDE_POSTS_ROW_COUNT  |
> +---------------------+----------------+-----------------+--------------------+--------------------------+------------------------+
> | PERFORMANCE_200000  |                |                 | null               
> | 2016-02-19 01:02:50.706  | null                   |
> | PERFORMANCE_200000  | STATS          | [B@21526f6c     | 5000063            
> |                          | 69100                  |
> | PERFORMANCE_200000  | STATS          | [B@49f5c307     | 5000010            
> |                          | 68975                  |
> | PERFORMANCE_200000  | USAGE          | [B@299266e2     | 5000060            
> |                          | 81022                  |
> | PERFORMANCE_200000  | USAGE          | [B@5471388b     | 5000064            
> |                          | 78177                  |
> | PERFORMANCE_200000  | USAGE          | [B@66ea1466     | 5000066            
> |                          | 80568                  |
> | PERFORMANCE_200000  | USAGE          | [B@1601e47      | 5000020            
> |                          | 78600                  |
> | PERFORMANCE_200000  | USAGE          | [B@3bffddff     | 5000019            
> |                          | 80377                  |
> | PERFORMANCE_200000  | USAGE          | [B@66971f6b     | 5000016            
> |                          | 78591                  |
> +---------------------+----------------+-----------------+--------------------+--------------------------+------------------------+
> {code}



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

Reply via email to