flink 1.8 on yarn，报RM失败

[九思]

请教各位老师：
   flink1.8 on yarn，yarn的ResourceManager使用的是HA模式，RM有2个节点（Active主节点是正常的，Standby备用节点没开）。提交任务会报错，连接standby节点的8032失败。为啥呢？既然是备用RM，为啥会访问它？
PS：flink1.7 on yarn，同样的yarn和hadoop环境，是可以提交的。


2019-07-23 09:28:25,621 INFO  org.apache.hadoop.io.retry.RetryInvocationHandler             - Exception while invoking ApplicationClientProtocolPBClientImpl.forceKillApplication over rm2 after 1 failover attempts. Trying to failover after sleeping for 21037ms.
java.net.ConnectException: Call From master/10.0.2.11 to slave1:8032 failed on connection exception: java.net.ConnectException: Connection refused; For more details see:  http://wiki.apache.org/hadoop/ConnectionRefused