近期Spark作业出现Yarn资源不足执行失败的问题,所以边学习边查看相关的Cloudera Manager配置。特此记录一下重点的关于资源方面的配置事项。
- Yarn 相关的性能配置
- yarn.nodemanager.resource.cpu-vcores: Container Virtual CPU Cores
- yarn.nodemanager.resource.memory-mb: Container Memory
- yarn.nodemanager.resource.percentage-physical-cpu-limit: Amount of CPU reserved for all the containers on each node
- yarn.scheduler.maximum-allocation-mb:The largest amount of physical memory, in MiB, that can be requested for a container.
- yarn.scheduler.maximum-allocation-vcores:The largest number of virtual CPU cores that can be requested for a container. This parameter has no effect prior to CDH 4.4.
- yarn.scheduler.fair.assignmultiple
- …
- ref: https://docs.cloudera.com/documentation/enterprise/6/properties/6.3/topics/cm_props_cdh5160_yarn_mr2included_.html#concept_6.3.x_gateway_props__section_performance_props
- Impala
- mem_limit:Impala Daemon Memory Limit