bug: groupby result not merged in cluster mode

[FANNG1]

Search before asking

• I had searched in the issues and found no similar issues.

Version

xx

What's Wrong?

run a sql in 3 nodes cluster, generates result like

20221010, 10
20221010, 20
20221010, 30

while the right reulst is 20221010, 60, seems we distribute data by p_date and author_id, and should add a final merge in root fragment

sql

SELECT a.d_f_date,
       sum(a.m_live_pay_unrisk_order_amt) AS sqlrewriter0
FROM
  (SELECT `p_date` AS `d_f_date`,
          author_id,
          coalesce(SUM(`pay_unrisk_order_amt`), 0) AS `m_live_pay_unrisk_order_amt`
   FROM tableA
   WHERE p_date BETWEEN '20211001' AND '20211115'
     AND pay_unrisk_order_amt < 2000000
     AND pay_unrisk_order_amt > 0
   GROUP BY `p_date`,
            author_id) AS a
GROUP BY a.d_f_date
ORDER BY a.d_f_date,
         sqlrewriter0
LIMIT 100
;

Fragment 0:
  DataExchange: Shuffle
  Exchange Sink: fragment id: [1]
    Aggregate(Partial): group items: [199, 0], aggregate functions: [SUM(179)]
      Filter: [>=($1, 20211001), <=($1, 20211115), <($2, 2000000), >($2, 0)]
        TableScan: [tableA]

Fragment 1:
  DataExchange: Merge
  Exchange Sink: fragment id: [2]
    EvalScalar: [$0]
      Aggregate(Final): group items: [199], aggregate functions: [sum(200)]
        Aggregate(Partial): group items: [199], aggregate functions: [sum(200)]
          EvalScalar: [multi_if(is_not_null($0), assume_not_null($0), true, 0, NULL)]
            Aggregate(Final): group items: [199, 0], aggregate functions: [SUM(179)]
              Exchange Source: fragment id: [0]

Fragment 2:
  Limit: [100], Offset: [0]
    Sort: [199 ASC, 204 ASC], Limit: [100]
      Exchange Source: fragment id: [1]

How to Reproduce?

No response

Are you willing to submit PR?

• Yes I am willing to submit a PR!

[sundy-li]

Related to distributed pipeline builder. cc @leiysky

[FANNG1]

@leiysky , could you explain how u fixed #8290 and #8292 ?