Zhang Mingli

Rebased and fixed: Bump catversion in dadeb84c0266d0117fd77cfa5ab8719d19b878a3

Update: two commits added to resolve the missed SCHEDULE clause when pg_dump DYNAMIC TABLE. b215893a5dc367462d956a5113c2723cb090a8c5 Add it when a Dynamic Table is dumped. Dynamic Table's SCHEDULE clause is stored in...

> @avamingli Could help verify the issue. IIUC, it's not libpq connection but network connections of OS, more processes could consume more interconnect connections. You could set lower parallel degree...

> max_parallel_workers_per_gather=64 is also in greenplum . Greenplum doesn't have parallel plan feature, it's a dead GUC for GPDB. It doesn't matter what you set it in GPDB. CBDB parallel...

> solution > > pip3 uninstall pg pip3 uninstall PyGreSQL > > then pip3 install -U PyGreSQL Thanks @yihong0618 . Generally, I think we need to update XXX.dependency if XXX...

> @tuhaihe @avamingli @Zhangbaowen-Hashdata We need to use `pip3 install -r ../../python-dependencies.txt` like [Readme.CentOS.bash.](https://github.com/cloudberrydb/cloudberrydb/blob/main/deploy/build/README.CentOS.bash#L34) In this way, Python dependencies will use a fixed version instead of relying on the version...

Do we have reproduce SQL steps?

Hi all, I try to reproduce with some data, but got the opposite result: with_cte is slower than normal agg. Here is my case, I use AO table to make...

Same for ORCA: ```sql explain(analyze) select count(distinct id) from ao ; QUERY PLAN ------------------------------------------------------------------------------------------------------------------------------- --- Finalize Aggregate (cost=0.00..440.60 rows=1 width=8) (actual time=326.314..326.317 rows=1 loops=1) -> Gather Motion 3:1 (slice1; segments:...

> I think you had reproduced the problem: we generated two different plans for those two SQL statements which are functionally equivalent. One uses the GroupAggregate node to perform deduplication,...