Spark资源配置（核数与内存）_如何在spark监控页面查看任务占用的内存和cpu

 

关于所在节点核数怎么看？

======================================================================

# 总核数 = 物理CPU个数 X 每颗物理CPU的核数

# 总逻辑CPU数 = 物理CPU个数 X 每颗物理CPU的核数 X 超线程数

 

# 查看物理CPU个数

cat /proc/cpuinfo| grep "physical id"| sort| uniq| wc -l

 

# 查看每个物理CPU中core的个数(即核数)

cat /proc/cpuinfo| grep "cpu cores"| uniq

 

# 查看逻辑CPU的个数

cat /proc/cpuinfo| grep "processor"| wc -l

======================================================================

 

spark资源主要就是core和memery。

 

spark主题功能分三部分：spark RDD，sparkSQL，spark shell，如果每个部分的功能都要用，那么每块都要占用资源。

 

其中，spark RDD和spark shell 是动态分配占用资源的，sparkSQL是静态分配资源的（启动后即一直占着分配的资源）

 

spark分配的总体资源在哪里看？

cat /home/mr/spark/conf/spark-env.sh

JAVA_HOME=/usr/java/jdk

SPARK_HOME=/home/mr/spark

SPARK_PID_DIR=/home/mr/spark/pids

SPARK_LOCAL_DIRS=/data2/zdh/spark/tmp,/data3/zdh/spark/tmp,/data4/zdh/spark/tmp

SPARK_WORKER_DIR=/data2/zdh/spark/work

SPARK_LOG_DIR=/data1/zdh/spark/logs

SPARK_HISTORY_OPTS="-Dspark.history.ui.port=18088-Dspark.history.retainedApplications=500"

SPARK_MASTER_WEBUI_PORT=18080

SPARK_WORKER_WEBUI_PORT=18081

SPARK_WORKER_CORES=25

SPARK_WORKER_MEMORY=150g

SPARK_DAEMON_MEMORY=2g

SPARK_LOCAL_HOSTNAME=`hostname`

YARN_CONF_DIR=/home/mr/yarn/etc/hadoop

SparkSQL的总体资源在哪看？

cat /home/mr/spark/conf/sparksql-default.conf

JAVA_HOME=/usr/java/jdk

SPARK_HOME=/home/mr/spark

SPARK_PID_DIR=/home/mr/spark/pids

SPARK_LOCAL_DIRS=/data2/zdh/spark/tmp,/data3/zdh/spark/tmp,/data4/zdh/spark/tmp

SPARK_WORKER_DIR=/data2/zdh/spark/work

SPARK_LOG_DIR=/data1/zdh/spark/logs

SPARK_HISTORY_OPTS="-Dspark.history.ui.port=18088-Dspark.history.retainedApplications=500"

SPARK_MASTER_WEBUI_PORT=18080

SPARK_WORKER_WEBUI_PORT=18081

SPARK_WORKER_CORES=25

SPARK_WORKER_MEMORY=150g

SPARK_DAEMON_MEMORY=2g

SPARK_LOCAL_HOSTNAME=`hostname`

YARN_CONF_DIR=/home/mr/yarn/etc/hadoop

[root@vmax47 conf]# catsparksql-defaults.conf

spark.serializer=org.apache.spark.serializer.KryoSerializer

spark.driver.extraJavaOptions=-Xss32m-XX:PermSize=128M -XX:MaxPermSize=512m

spark.driver.extraClassPath=/home/mr/spark/libext/*

spark.executor.extraClassPath=/home/mr/spark/libext/*

spark.executor.memory=10g

spark.eventLog.enabled=true

spark.eventLog.dir=/data1/zdh/spark/logs/eventLog

spark.history.fs.logDirectory=/data1/zdh/spark/logs/eventLog

spark.worker.cleanup.enabled=true

spark.shuffle.consolidateFiles=true

spark.ui.retainedJobs=200

spark.ui.retainedStages=200

spark.deploy.retainedApplications=100

spark.deploy.retainedDrivers=100

spark.speculation=true

spark.speculation.interval=1000

spark.speculation.multiplier=4

spark.speculation.quantile=0.85

spark.shuffle.service.enabled=false

spark.dynamicAllocation.enabled=false

spark.dynamicAllocation.minExecutors=0

spark.dynamicAllocation.maxExecutors=2147483647

spark.sql.broadcastTimeout=600

spark.yarn.queue=mr

spark.master=spark://vmax47:7077,SPARK49:7077

spark.deploy.recoveryMode=ZOOKEEPER

spark.deploy.zookeeper.url=SPARK49:2181,HADOOP50:2181,vmax47:2181

spark.ui.port=4100

spark.driver.memory=40G

spark.cores.max=30

 

 

查看Spark资源可从18080端口查看：