You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Eugene Koifman (JIRA)" <ji...@apache.org> on 2016/10/24 20:06:58 UTC

[jira] [Created] (HIVE-15044) LockManager may be too coarse grained

Eugene Koifman created HIVE-15044:
-------------------------------------

             Summary: LockManager may be too coarse grained 
                 Key: HIVE-15044
                 URL: https://issues.apache.org/jira/browse/HIVE-15044
             Project: Hive
          Issue Type: Improvement
          Components: Transactions
    Affects Versions: 1.0.0
            Reporter: Eugene Koifman
            Assignee: Eugene Koifman


Consider
{noformat}
create table target (a int, b int)
      partitioned by (p int, q int) clustered by (a) into 2  buckets 
      stored as orc TBLPROPERTIES ('transactional'='true')")
insert into target partition(p=1,q) values (1,2,3)
{noformat}

this insert will lock the whole table.  See 
{noformat}
DbTxnManager.acquireLocks()
switch (output.getType()) {
        case DUMMYPARTITION:   //
{noformat}

Insert operation runs with SHARED_READ lock but once HIVE-15032 is addressed this will be an issue for Update/Delete/Merge which use a more restrictive SHARED_WRITE lock.



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