hive-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Eugene Koifman (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HIVE-13470) Too many locks acquired for partitioned read
Date Thu, 26 May 2016 17:33:12 GMT

     [ https://issues.apache.org/jira/browse/HIVE-13470?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Eugene Koifman updated HIVE-13470:
----------------------------------
    Target Version/s: 1.3.0, 2.2.0  (was: 1.3.0, 2.1.0)

> Too many locks acquired for partitioned read
> --------------------------------------------
>
>                 Key: HIVE-13470
>                 URL: https://issues.apache.org/jira/browse/HIVE-13470
>             Project: Hive
>          Issue Type: Bug
>          Components: Transactions
>    Affects Versions: 1.3.0
>            Reporter: Eugene Koifman
>            Assignee: Eugene Koifman
>
> consider 
> {noformat}
> create table TAB_PART (a int, b int) partitioned by (p string) clustered by (a) into
2  buckets stored as orc TBLPROPERTIES ('transactional'='true')
> select a from  TAB_PART where p = 'blah'
> {noformat}
> If the table is truly empty (exactly as above) then DbLockManger will acquire SHARED_READ
lock on the table (for the select stmt)
> If prior to Select, one runs "alter table TAB_PART add partition (p = 'blah')" then 2
SHARED_LOCKS are acquired: 1 on table and 1 on partition.
> Should only get 1 partition level lock.
> The behavior of lock manager is such because the generated query plan creates ReadEntity
objects this way.
> Todo: try actually inserting data and see if that changes anything.



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

Mime
View raw message