HIVE-28121. (2.3) Use direct SQL for transactional altering table parameter #5204
Conversation
| @Override | ||
| public long updateParameterWithExpectedValue(Table table, String key, String expectedValue, String newValue) | ||
| throws MetaException, NoSuchObjectException { | ||
| final Table _table = table; |
There was a problem hiding this comment.
this might be limitation of Java 7
asf-ci-hive
added
tests failed
tests pending
and removed
tests pending
tests failed
labels
lirui-apache
left a comment
lirui-apache
left a comment
There was a problem hiding this comment.
+1, thanks @pan3793 for taking care of this
|
@pvary Could you please take a look at this one? AFAIK it's the last one before the next RC for 2.3.10 |
|
kindly ping @sunchao, are we good to go? |
|
Thanks @pan3793 @lirui-apache and @pvary . It looks like this PR has more tests failing than the previous PR: 
In the previous PR: #4892, there were only 27 tests failing (normally there would only be 26 tests but 1 was flaky in that PR). Could you double check whether the test failures are related? |
Yes that's unfortunately the current state of branch-2.3. We do know a fixed set of tests are failing though and are using that as the baseline for testing against PRs. |
|
@sunchao based on the CI results of this PR branch, I think the failure tests should be flaky cases |
|
OK, just re-triggered CI to confirm. |
asf-ci-hive
added
tests failed
tests pending
and removed
tests pending
tests failed
labels
|
Hmm @pan3793 I tried the tests again for 2 times, and each time there were over 40 tests failing, which is quite more than the normal 26. Could you double-check whether there is anything we should fix here? Thanks. |
asf-ci-hive
added
tests failed
tests pending
and removed
tests pending
tests failed
labels
|
oops, thanks @pan3793 - let me check the commits in branch-2.3 and see which commit caused the regression |
|
Merged, thanks |
|
@sunchao can we have the next RC for 2.3.10? |
|
Sorry for the delay @pan3793 . I just sent out RC1. |
|
|
||
| long updateTableParam(Table table, String key, String expectedValue, String newValue) { | ||
| String queryText = String.format("UPDATE \"TABLE_PARAMS\" SET \"PARAM_VALUE\" = ? " + | ||
| "WHERE \"PARAM_KEY\" = ? AND \"PARAM_VALUE\" = ? AND \"TBL_ID\" IN " + |
There was a problem hiding this comment.
Just for reference, we hit performance issues in our production environment on executing this SQL. Changing the IN to = would solve this issue, at least for MySQL 5.7
There was a problem hiding this comment.
Thanks @pan3793 for reporting the issue. I agree it's better to use scalar subquery here. But since both 2.x and 3.x lines are EOL, and subquery is not used in 4.0, so I guess we won't have to do anything at the moment, right?
There was a problem hiding this comment.
Yes, the Hive 4.0 doesn't have such an issue. This is just a tip for users who want to port this patch to their internal legacy Hive version (2.3 or 3.x)
5 participants
What changes were proposed in this pull request?
Port #5129 to branch-2.3
Why are the changes needed?
This is a follow-up of HIVE-26882, including correctness fixes for some RDBMS.
Does this PR introduce any user-facing change?
No
Is the change a dependency upgrade?
No
How was this patch tested?
UT