Unexpected results in intermediate table when downsampling two tables' data on a monthly basis through stream processing

[nieyuan1980]

一、对两个表具按月取最大值进行降采样

select _wstart recode_time, meter_code, max(dqljlcc) max_value
from water_meters1
where meter_code in ('METXJJSZXNSB1', 'METXJNGWDXFSX')
partition by meter_code, location, working_on, is_calc interval (1n)
order by meter_code, recode_time;

二、使用流式计算将结果写入中间表

create stream water_meter_ns trigger at_once fill_history 1
into water_meter_n
    tags(meter_code varchar(32), location varchar(512), working_on varchar(32), is_calc varchar(8))
    subtable(cast(concat('n_', lower(meter_code)) as varchar(32))) as
select _wstart recode_time, max (dqljlcc) max_value
from water_meters1
where meter_code in ('METXJJSZXNSB1', 'METXJNGWDXFSX')
partition by meter_code, location, working_on, is_calc interval (1n);

三、查询中间表

select * from order by meter_code, recode_time;

• OS: [CentOS 7.0]
• CPU: [CentOS 7.0],
• Memory: [2G]
• Disk Space: [40G]
• TDengine Version [3.2.0.0]