为了更快的定位您的问题，请提供以下信息，谢谢
【详述】物化视图改写失败
【背景】做过哪些操作？
【业务影响】测试中，无影响
【是否存算分离】否
【StarRocks版本】例如：3.2.3
【集群规模】例如：3fe（1 follower+2observer）+3be（fe与be混部）
【机器信息】CPU虚拟核/内存/网卡，例如：48C/64G/万兆
【联系方式】为了在解决问题过程中能及时联系到您获取一些日志信息，请补充下您的联系方式，例如：社区群4-小李或者邮箱，谢谢
【附件】
异步物化视图
CREATE MATERIALIZED VIEW test.project_record_area_code_base
DISTRIBUTED BY HASH(cdate)
PROPERTIES (
‘query_rewrite_consistency’=‘LOOSE’
) as
SELECT
area_code AS areacode,
COUNT(called_num) AS monthCount1,
COUNT(DISTINCT called_num) AS monthCount2,
COUNT(CASE WHEN call_duration > 0 THEN called_num ELSE NULL END),
COUNT(CASE WHEN call_duration >= 20 THEN called_num ELSE NULL END) AS monthCount4,
COUNT(DISTINCT CASE WHEN call_duration > 0 THEN called_num ELSE NULL END) AS monthCount5,
COUNT(DISTINCT CASE WHEN call_duration >= 20 THEN called_num ELSE NULL END) AS monthCount6,
COUNT(CASE WHEN end_result = 4 THEN called_num ELSE NULL END) AS monthCount7,
COUNT(DISTINCT CASE WHEN sms_content LIKE CONCAT(’%’,8,’%’) THEN called_num ELSE NULL END) AS monthCount10,
cdate
FROM project_record_dayPartitionTest record
LEFT JOIN project p ON p.id = project_id
WHERE cdate <= date(date_sub(current_timestamp(),1))
GROUP BY area_code,cdate;
sql1：
explain SELECT
area_code AS areacode,
COUNT(called_num) AS monthCount1,
COUNT(DISTINCT called_num) AS monthCount2,
cdate
FROM project_record_dayPartitionTest record
LEFT JOIN project p ON p.id = project_id
WHERE cdate <= date(date_sub(current_timestamp(),0))
GROUP BY area_code,cdate;
可以触发查询改写
explain1.data (4.1 KB)
sql2
explain SELECT
area_code AS areacode,
COUNT(called_num) AS monthCount1,
COUNT(DISTINCT called_num) AS monthCount2,
COUNT(CASE WHEN call_duration > 0 THEN called_num ELSE NULL END),
cdate
FROM project_record_dayPartitionTest record
LEFT JOIN project p ON p.id = project_id
WHERE cdate <= date(date_sub(current_timestamp(),0))
GROUP BY area_code,cdate;
无法触发查询改写
sql2只是比sql1 多了个COUNT(CASE WHEN call_duration > 0 THEN called_num ELSE NULL END), 这个语句在异步物化视图里面有的
explain.data (1.9 KB)
- fe.log/beINFO/相应截图
- 慢查询：
- Profile信息，获取Profile，通过Profile分析查询瓶颈
- 并行度：show variables like ‘%parallel_fragment_exec_instance_num%’;
- pipeline是否开启：show variables like ‘%pipeline%’;
- be节点cpu和内存使用率截图
- 查询报错：
- query_dump，怎么获取query_dump文件
- be crash
- be.out
- coredump，如何获取coredump
- 外表查询报错
- be.out和fe.warn.log
