Bug: create table xxx engine = memory as xxx lost columns under cluster

[xudong963]

Summary

Tables in select are from ci tpcds dataset,

create or replace table test engine =memory as SELECT sr_customer_sk AS ctr_customer_sk,           sr_store_sk AS ctr_store_sk,           sum(sr_return_amt) AS ctr_total_return    FROM store_returns,       date_dim    WHERE sr_returned_date_sk = d_date_sk      AND d_year = 2000    GROUP BY sr_customer_sk,             sr_store_sk;

After creating test, it's ok to query the count, but will report error if query the specific data.

mysql> select count() from test;
+---------+
| count() |
+---------+
|     364 |
+---------+
1 row in set (0.06 sec)
Read 364 rows, 0.00 B in 0.012 sec., 30.57 thousand rows/sec., 0.00 B/sec.

mysql> select * from test;
ERROR 1105 (HY000): PanicError. Code: 1104, Text = index out of bounds: the len is 0 but the index is 0.

After some investigations, I found that some data blocks written to the memory table test don't have columns(zero col but non-zero rows).

[xudong963]

The fuse engine is okay, and the single machine is okay. So, I guess there are some issues with the memory table and distribution. Any thoughts? @zhyass @dantengsky @zhang2014

[sundy-li]

Memory engines don't support distribution I think.

Data is not shared.