flink启动作业内存优化

一、The configuration value ‘containerized.heap-cutoff-min’ =‘1024’ is larger than the total container memory 64 或者 The configuration value ‘containerized.heap-cutoff-min’ is higher (600) than the requested amount of memory 512

源码：ContaineredTaskManagerParameters.calculateCutoffMB()方法
flink-conf.yaml配置：

containerized.heap-cutoff-min: 512

表示 承载 TaskManager 的 YARN Container 需要预留512M内存给TaskManager 之外的逻辑来使用(该参数默认为600M)

如果jm和tm参数设为64，明显没有额外的内存预留，启动不了yarn session

二、Invalid configuration value for (taskmanager.memory.network.fraction,taskmanager.memory.network.min,taskmanager.memory.network.max) : (0.1, 536870912, 536870912) - Network buffer memory size to large: 536870912 >= 92274688 (total JVM memory size)

taskmanager.memory.network.max: 512mb
taskmanager.memory.network.min: 512mb

如果jm和tm参数设为600m，network占用512m，JVM仅有88m，肯定Could not deploy Yarn job cluster

tm参数至少为jm + containerized.heap-cutoff-min+taskmanager.network.memory.min