maxContainerCapability 设置不足

异常：

REDUCE capability required is more than the supported max container capability in the cluster. Killing the Job. reduceResourceRequest: <memory:12160, vCores:1> maxContainerCapability:<memory:8192, vCores:24>

 

需要调整两个参数：

yarn.nodemanager.resource.memory-mb

yarn.scheduler.maximum-allocation-mb

 

注意：如果只设置其中一个，可能依然存在同样的报错