gianm commented on PR #12521:
URL: https://github.com/apache/druid/pull/12521#issuecomment-1130240775

   I've run CI a bunch of times on this patch and haven't seen issues with 
buffer allocations, which is good. But I'm having problems with other flaky 
tests that I am having a tough time reproducing locally. I'm wondering if 
anyone has any insights? Here's the two that are popping up repeatedly.
   
   In https://app.travis-ci.com/github/apache/druid/jobs/570542846 (and other 
places):
   
   ```
   GroupByQueryRunnerTest.testGroupByWithExpressionAggregatorWithComplex:12641 
subquery-cardinality-0: idx[1] expected:<9.019833517963864> but was:<0.0>
   ```
   
   `idx[1]` is `carExpr`, an ExpressionLambdaAggregatorFactory using 
`hyper_unique`, `hyper_unique_add(quality, __acc)`, and 
`hyper_unique_estimate(o)`. `idx[0]`, which must be OK (since the checker got 
to `idx[1]`), is `car`, a CardinalityAggregatorFactory on `quality`.
   
   In https://app.travis-ci.com/github/apache/druid/jobs/570542848 (and other 
places):
   
   ```
   ScanQueryRunnerTest.testFullOnSelectAsCompactedList:255->verify:1169 invalid 
null value for doubleNumericNull (expected 0.000000)
   ```
   
   The linked job is run with SQL Compatibility. Strangely, when I run the test 
in this mode, the expected value for `doubleNumericNull` is never `0`. It's 
either 10, 20, or `null`. So perhaps the expected value is wrong for some 
reason.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to