Vertica 更新出现重新分段问题

Question

    UPDATE
        target_table AS tgt 
        SET
        campaign_name = src.campaign_name,
            line_of_business = src.Campaign_Line_Of_Business
        FROM
        (
        SELECT
            A.project_id ,
            A.campaign_id ,
            A.run_id ,
            B.Campaign_Line_Of_Business,
            B.campaign_name
        FROM
            target_table A
        left JOIN src_table B 
                 ON
            (A.Project_id = B.Project_id
                AND A.Campaign_id = B.Campaign_id
                AND A.Run_id = B.Run_id)
        GROUP BY
            1,
            2,
            3,
            4,
            5 
        ) src
        WHERE
        (tgt.project_id = src.project_id
            and tgt.campaign_id = src.campaign_id
            and tgt.run_id = src.run_id);

投影分割和排序...

        ORDER BY target_table.project_id,
              target_table.campaign_id,
              target_table.run_id,
              target_table.campaign_node_id,
              target_table.locale_cd
        SEGMENTED BY hash(target_table.project_id, 
        target_table.campaign_id, 
        target_table.run_id, 
        target_table.campaign_node_id, 
        target_table.message_id, 
        target_table.locale_cd) ALL NODES KSAFE 1;

Answer 1

我不完全确定 - 当您在 UPDATE 语句的 FROM 子句中使用由 join 和 group by 组成的查询时，这是一种特殊情况。

但是：

驱动输入表 - 也是目标表，由这六个的哈希分段：

SEGMENTED BY hash(
        target_table.project_id, 
        target_table.campaign_id, 
        target_table.run_id, 
        target_table.campaign_node_id, 
        target_table.message_id, 
        target_table.locale_cd
)

...连接 WHERE 条件由以下 3 个组成：

    tgt.project_id  = src.project_id
and tgt.campaign_id = src.campaign_id
and tgt.run_id      = src.run_id

如果我们按列

a,b,c,d,e,f

进行分段，并且这些都是整数，则组合：

a | b | c | d | e | f
1 | 1 | 1 | 1 | 1 | 1

可以在节点 1 上，并且组合：

a | b | c | d | e | f
1 | 1 | 1 | 1 | 2 | 1

，因为它会导致不同的哈希值，因此可能位于节点 2 上。 因此用于连接的

a,b,c

组合将导致所有节点上都存在相同的组合，因此重新分段对于使连接成为可能是必不可少的。