创建zybo cluster的spark集群（计算层面）：
1.每个节点都是同样的filesystem，mac地址冲突，故：
vi ./etc/profile
export PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:$PATH
export JAVA_HOME=/usr/lib/jdk1.7.0_55
export CLASSPATH=.:$JAVA_HOME/lib/tools.jar
export PATH=$JAVA_HOME/bin:$PATH
export HADOOP_HOME=/root/hadoop-2.4.0
ifconfig eth1 hw ether 00:0a:35:00:01:03
ifconfig eth1 192.168.1.3/24 up
2.生成私匙 id_rsa 与 公匙 id_rsa.pub 配置文件
ssh-keygen -t rsa
id_rsa是密钥文件，id_rsa.pub是公钥文件。
3.Worker节点/etc/hosts配置：
具体操作步骤：
ssh root@192.168.1.x
vi /etc/hosts
127.0.0.1 localhost zynq
192.168.1.1 spark1
192.168.1.2 spark2
192.168.1.3 spark3
192.168.1.4 spark4
192.168.1.5 spark5
192.168.1.100 sparkMaster
#::1 localhost ip6-localhost ip6-loopback
Master节点/etc/hosts配置：
4.分发公钥
ssh-copy-id -i ~/.ssh/id_rsa.pub root@spark1
ssh-copy-id -i ~/.ssh/id_rsa.pub root@spark2
ssh-copy-id -i ~/.ssh/id_rsa.pub root@spark3
ssh-copy-id -i ~/.ssh/id_rsa.pub root@spark4
…..
5.配置Master节点
Cd ~/spark-0.9.1-bin-hadoop2/conf
Vi slaves
6.配置java
否则运行pi计算时会出现count找不到的错误（因为pyspark找不到javaruntime）。
cd /usr/bin/
ln -s /usr/lib/jdk1.7.0_55/bin/java java
ln -s /usr/lib/jdk1.7.0_55/bin/javac javac
ln -s /usr/lib/jdk1.7.0_55/bin/jar jar
7.测试运行所有节点
SPARK_MASTER_IP=192.168.1.1 ./sbin/start-all.sh
SPARK_MASTER_IP=192.168.1.100 ./sbin/start-all.sh
成功启动所有节点：
8.查看工作状态：
Jps
Netstat -ntlp
9.开启脚本命令行
MASTER=spark://192.168.1.1:7077 ./bin/pyspark
MASTER=spark://192.168.1.100:7077 ./bin/pyspark
10.测试
from random import random
def sample(p):
x, y = random(), random()
return 1 if x*x + y*y < 1 else 0
count = sc.parallelize(xrange(0, 1000000)).map(sample) \
.reduce(lambda a, b: a + b)
print "Pi is roughly %f" % (4.0 * count / 1000000)
成功进行运算：
正常启动信息：
root@zynq:~/spark-0.9.1-bin-hadoop2# MASTER=spark://192.168.1.1:7077 ./bin/pyspark
Python 2.7.4 (default, Apr 19 2013, 19:49:55)
[GCC 4.7.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.
log4j:WARN No appenders could be found for logger (akka.event.slf4j.Slf4jLogger).
log4j:WARN Please initialize the log4j system properly.
log4j:WARN See http://logging.apache.org/log4j/1.2/faq.html#noconfig for more info.
70/01/01 00:07:48 INFO SparkEnv: Using Spark‘s default log4j profile: org/apache/spark/log4j-defaults.properties
70/01/01 00:07:48 INFO SparkEnv: Registering BlockManagerMaster
70/01/01 00:07:49 INFO DiskBlockManager: Created local directory at /tmp/spark-local-19700101000749-e1fb
70/01/01 00:07:49 INFO MemoryStore: MemoryStore started with capacity 297.0 MB.
70/01/01 00:07:49 INFO ConnectionManager: Bound socket to port 36414 with id = ConnectionManagerId(spark1,36414)
70/01/01 00:07:49 INFO BlockManagerMaster: Trying to register BlockManager
70/01/01 00:07:49 INFO BlockManagerMasterActor$BlockManagerInfo: Registering block manager spark1:36414 with 297.0 MB RAM
70/01/01 00:07:49 INFO BlockManagerMaster: Registered BlockManager
70/01/01 00:07:49 INFO HttpServer: Starting HTTP Server
70/01/01 00:07:50 INFO HttpBroadcast: Broadcast server started at http://192.168.1.1:42068
70/01/01 00:07:50 INFO SparkEnv: Registering MapOutputTracker
70/01/01 00:07:50 INFO HttpFileServer: HTTP File server directory is /tmp/spark-77996902-7ea4-4161-bc23-9f3538967c17
70/01/01 00:07:50 INFO HttpServer: Starting HTTP Server
70/01/01 00:07:51 INFO SparkUI: Started Spark Web UI at http://spark1:4040
70/01/01 00:07:52 INFO AppClient$ClientActor: Connecting to master spark://192.168.1.1:7077...
70/01/01 00:07:55 INFO SparkDeploySchedulerBackend: Connected to Spark cluster with app ID app-19700101000755-0001
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor added: app-19700101000755-0001/0 on worker-19700101000249-spark2-53901 (spark2:53901) with 2 cores
70/01/01 00:07:55 INFO SparkDeploySchedulerBackend: Granted executor ID app-19700101000755-0001/0 on hostPort spark2:53901 with 2 cores, 512.0 MB RAM
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor added: app-19700101000755-0001/1 on worker-19700101000306-spark5-38532 (spark5:38532) with 2 cores
70/01/01 00:07:55 INFO SparkDeploySchedulerBackend: Granted executor ID app-19700101000755-0001/1 on hostPort spark5:38532 with 2 cores, 512.0 MB RAM
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor added: app-19700101000755-0001/2 on worker-19700101000255-spark3-41536 (spark3:41536) with 2 cores
70/01/01 00:07:55 INFO SparkDeploySchedulerBackend: Granted executor ID app-19700101000755-0001/2 on hostPort spark3:41536 with 2 cores, 512.0 MB RAM
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor added: app-19700101000755-0001/3 on worker-19700101000254-spark4-38766 (spark4:38766) with 2 cores
70/01/01 00:07:55 INFO SparkDeploySchedulerBackend: Granted executor ID app-19700101000755-0001/3 on hostPort spark4:38766 with 2 cores, 512.0 MB RAM
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor updated: app-19700101000755-0001/0 is now RUNNING
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor updated: app-19700101000755-0001/3 is now RUNNING
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor updated: app-19700101000755-0001/1 is now RUNNING
70/01/01 00:07:55 INFO AppClient$ClientActor: Executor updated: app-19700101000755-0001/2 is now RUNNING
70/01/01 00:07:56 WARN NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Welcome to
____ __
/ __/__ ___ _____/ /__
_\ \/ _ \/ _ `/ __/ ‘_/
/__ / .__/\_,_/_/ /_/\_\ version 0.9.1
/_/
Using Python version 2.7.4 (default, Apr 19 2013 19:49:55)
Spark context available as sc.
>>> 70/01/01 00:08:06 INFO SparkDeploySchedulerBackend: Registered executor: Actor[akka.tcp://sparkExecutor@spark3:35842/user/Executor#1876589543] with ID 2
70/01/01 00:08:11 INFO BlockManagerMasterActor$BlockManagerInfo: Registering block manager spark3:42847 with 297.0 MB RAM
70/01/01 00:08:12 INFO SparkDeploySchedulerBackend: Registered executor: Actor[akka.tcp://sparkExecutor@spark5:43445/user/Executor#-1199017431] with ID 1
70/01/01 00:08:13 INFO BlockManagerMasterActor$BlockManagerInfo: Registering block manager spark5:42630 with 297.0 MB RAM
70/01/01 00:08:15 INFO AppClient$ClientActor: Executor updated: app-19700101000755-0001/0 is now FAILED (Command exited with code 1)
70/01/01 00:08:15 INFO SparkDeploySchedulerBackend: Executor app-19700101000755-0001/0 removed: Command exited with code 1
70/01/01 00:08:15 INFO AppClient$ClientActor: Executor added: app-19700101000755-0001/4 on worker-19700101000249-spark2-53901 (spark2:53901) with 2 cores
70/01/01 00:08:15 INFO SparkDeploySchedulerBackend: Granted executor ID app-19700101000755-0001/4 on hostPort spark2:53901 with 2 cores, 512.0 MB RAM
70/01/01 00:08:15 INFO AppClient$ClientActor: Executor updated: app-19700101000755-0001/4 is now RUNNING
70/01/01 00:08:21 INFO SparkDeploySchedulerBackend: Registered executor: Actor[akka.tcp://sparkExecutor@spark4:41692/user/Executor#-1994427913] with ID 3
70/01/01 00:08:26 INFO BlockManagerMasterActor$BlockManagerInfo: Registering block manager spark4:49788 with 297.0 MB RAM
70/01/01 00:08:27 INFO SparkDeploySchedulerBackend: Registered executor: Actor[akka.tcp://sparkExecutor@spark2:44449/user/Executor#-1155287434] with ID 4
70/01/01 00:08:28 INFO BlockManagerMasterActor$BlockManagerInfo: Registering block manager spark2:38675 with 297.0 MB RAM
原文：http://www.cnblogs.com/shenerguang/p/3822963.html