Xiaotian (Jackie) Jiang

When there is only one server queried, we can calculate the final aggregation result on the server side, instead of serializing back the intermediate result (which can be huge for...

Recover real-time consumption when consuming segment stuck

4

One possible solution is to manually delete the consuming segment and let the controller create a new one during real-time segment validation. But this scenario is not covered in the...

``` Error: RealtimeNonReplicaGroupSegmentAssignmentTest.testRelocateCompletedSegments:231 expected [true] but found [false] ``` It tests the behavior of Helix auto rebalance strategy (`SegmentAssignmentUtils.rebalanceTableWithHelixAutoRebalanceStrategy()`), which might not perfectly balance the segments.

Cut release-0.11.0 candidate branch again

1

Since we cut release-0.11.0 candidate branch, we have introduced several bug-fixes (list of new commits [here](https://github.com/apache/pinot/compare/release-0.11.0...master)), and several of them are critical because we recently upgraded several dependencies. I checked...

`RealtimeToOfflineSegmentsTask` does not honor `maxNumRecordsPerTask`, so it will generate only one sub-task per time bucket. In a table with high ingestion rate, each time bucket may contain >100M records, and...

In the instance config, entries with same key are treated as array which can cause unexpected behavior

2

E.g. In the controller config: ``` controller.enable.split.commit=true controller.enable.split.commit=true ``` will be treated as ``` controller.enable.split.commit=true,true ``` and then when pinot parses the value, `true,true` will be parsed as `false` This...

Enhance Data Ingestion Engine

3

Motivation: - Offline segment generation and real-time in-memory consuming to on-disk committed segment conversion are very similar process, and they are both achieved by the same segment creation engine. We...

Consider not performing alias rewrite to the rest of the query when alias is a column name

1

E.g. `SELECT SUM(a) AS a WHERE a > 10` will throw exception because it will be rewritten to `SELECT SUM(a) AS a WHERE SUM(A) > 10` Also, seems alias is...

Semantic of [ANY_VALUE](https://docs.snowflake.com/en/sql-reference/functions/any_value.html)

RangeBitmap can give wrong result when computing between on 2 values with different highest bit

2

**Describe the bug** When setting the RangeBitmap to full range (`0xFFFFFFFFFFFFFFFFL`, used for computing double range), computing a between for 2 values with different highest bit won't return the expected...