[
https://issues.apache.org/jira/browse/HIVE-14357?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15401653#comment-15401653
]
Rajat Khandelwal commented on HIVE-14357:
-----------------------------------------
+1. Changes look good.
One question: Do these tests fail in 2.1 release too? The commit seems to be a
part of 2.1.0-rc3.
Secondly, can we merge this to branch-2.1 soon? I'm kind of blocked on this for
something, Thanks.
> TestDbTxnManager2#testLocksInSubquery failing in branch-2.1
> -----------------------------------------------------------
>
> Key: HIVE-14357
> URL: https://issues.apache.org/jira/browse/HIVE-14357
> Project: Hive
> Issue Type: Bug
> Affects Versions: 2.1.1
> Reporter: Rajat Khandelwal
> Assignee: Sergey Shelukhin
> Attachments: HIVE-14357.patch
>
>
> {noformat}
> checkCmdOnDriver(driver.compileAndRespond("insert into R select * from S
> where a in (select a from T where b = 1)"));
> txnMgr.openTxn("three");
> txnMgr.acquireLocks(driver.getPlan(), ctx, "three");
> locks = getLocks();
> Assert.assertEquals("Unexpected lock count", 3, locks.size());
> checkLock(LockType.SHARED_READ, LockState.ACQUIRED, "default", "T", null,
> locks.get(0));
> checkLock(LockType.SHARED_READ, LockState.ACQUIRED, "default", "S", null,
> locks.get(1));
> checkLock(LockType.SHARED_READ, LockState.ACQUIRED, "default", "R", null,
> locks.get(2));
> {noformat}
> This test case is failing. The expected order of locks is supposed to be T,
> S, R. But upon closer inspection, it seems to be R,S,T.
> I'm not much familiar with what these locks are and why the order is
> important. Raising this jira so while I try to understand it all. Meanwhile,
> if somebody can explain here, would be helpful.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
