flink在upsert操作的时候, 是需要指定主键的, 否则会出现 requires that Table has a full primary keys if it is updated.
针对该问题, 给出的解决如下:

Exception in thread "main" org.apache.flink.table.api.TableException: UpsertStreamTableSink requires that Table has a full primary keys if it is updated.
	at org.apache.flink.table.planner.plan.nodes.physical.stream.StreamExecLegacySink.translateToPlanInternal(StreamExecLegacySink.scala:93)
	at org.apache.flink.table.planner.plan.nodes.physical.stream.StreamExecLegacySink.translateToPlanInternal(StreamExecLegacySink.scala:48)
	at org.apache.flink.table.planner.plan.nodes.exec.ExecNode$class.translateToPlan(ExecNode.scala:58)
	at org.apache.flink.table.planner.plan.nodes.physical.stream.StreamExecLegacySink.translateToPlan(StreamExecLegacySink.scala:48)
	at org.apache.flink.table.planner.delegation.StreamPlanner$$anonfun$translateToPlan$1.apply(StreamPlanner.scala:67)

flink1.11 后, 官方给出的解决思路是:

 create table t_sink_01 ( 
id varchar, 
name1 varchar,
name2 varchar,
ts timestamp(3),
money varchar,
primary key(id)  NOT ENFORCED  '这个是官方给出'
)
with( 
'connector.username' = '******' 
 ,'connector.type' = 'jdbc' 
 ,'connector.password' = '******' 
 ,'connector.table' = 't_student' 
 ,'connector.url' = 'jdbc:mysql://******:3306/******' 
 ,'connector.write.flush.max-rows' = '1'
) ;

'说明:'
primary key(id)  NOT ENFORCED  '意思是声明id为主键, 但是不做强检验!'

但是发现还是报同样的异常.
还是使用flink1.11之前的解决思路: 即套一个子查询, 进行group by 来让flink来推测出主键.

INSERT INTO t_snoring_flink_01 
select 
id ,
tsa, 
total_anti_snore_count,
total_anti_snore_during 
from (
select 
 CONCAT_WS('|',DATE_FORMAT(winTime, 'yyyy-MM-dd HH:mm:00'),deviceName ) as id,
 MAX(winTime) AS tsa,
 MAX(CAST(totalAntiSnoreCount AS integer)) AS total_anti_snore_count,
 MAX(CAST(totalAntiSnoreDuring AS BIGINT)) AS total_anti_snore_during
from  v_1144_4
group by CONCAT_WS('|',DATE_FORMAT(winTime, 'yyyy-MM-dd HH:mm:00'),deviceName )
) tt ;

我在这里使用是字符串拼接后生成的主键.