mesos集群部署参见上篇。
运行在mesos上面和 spark standalone模式的区别是：
1）stand alone
需要自己启动spark master
需要自己启动spark slaver（即工作的worker）
2）运行在mesos
启动mesos master
启动mesos slaver
启动spark的 ./sbin/start-mesos-dispatcher.sh -m mesos://127.0.0.1:5050
配置spark的可执行程序的路径(也就是mesos里面所谓EXECUTOR)，提供给mesos下载运行。
在mesos上面的运行流程：
1）通过spark-submit提交任务到spark-mesos-dispatcher
2）spark-mesos-dispatcher 把通过driver 提交到mesos master，并收到任务ID
3）mesos master 分配到slaver 让它执行任务
4) spark-mesos-dispatcher，通过任务ID查询任务状态
前期配置：
1) spark-env.sh
配置mesos库，以及spark可以执行的二进制程序包（这里偷懒用了官网的包，这里URI支持hdfs，http）。
#到spark的安装目录
vim conf/spark-env.sh # Options read by executors and drivers running inside the cluster
# - SPARK_LOCAL_IP, to set the IP address Spark binds to on this node
# - SPARK_PUBLIC_DNS, to set the public DNS name of the driver program
# - SPARK_CLASSPATH, default classpath entries to append
# - SPARK_LOCAL_DIRS, storage directories to use on this node for shuffle and RDD data
# - MESOS_NATIVE_JAVA_LIBRARY, to point to your libmesos.so if you use Mesos
MESOS_NATIVE_JAVA_LIBRARY=/usr/local/lib/libmesos.so
SPARK_EXECUTOR_URI=http://d3kbcqa49mib13.cloudfront.net/spark-1.5.1-bin-hadoop2.6.tgz
2）spark-defaults.conf
# Default system properties included when running spark-submit.
# This is useful for setting default environmental settings. # Example:
# spark.master spark://master:7077
# spark.eventLog.enabled true
# spark.eventLog.dir hdfs://namenode:8021/directory
# spark.serializer org.apache.spark.serializer.KryoSerializer
# spark.driver.memory 5g
# spark.executor.extraJavaOptions -XX:+PrintGCDetails -Dkey=value -Dnumbers="one two three" spark.executor.uri http://d3kbcqa49mib13.cloudfront.net/spark-1.5.1-bin-hadoop2.6.tgz
spark.master mesos://10.230.136.197:5050
3) 修改测试WordCount.java
这里测试例子，可以参考 提交任务到spark。
/**
* Illustrates a wordcount in Java
*/
package com.oreilly.learningsparkexamples.mini.java; import java.util.Arrays;
import java.util.List;
import java.lang.Iterable; import scala.Tuple2; import org.apache.commons.lang.StringUtils; import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaRDD;
import org.apache.spark.api.java.JavaPairRDD;
import org.apache.spark.api.java.JavaSparkContext;
import org.apache.spark.api.java.function.FlatMapFunction;
import org.apache.spark.api.java.function.Function2;
import org.apache.spark.api.java.function.PairFunction; public class WordCount {
public static void main(String[] args) throws Exception {
String inputFile = args[0];
String outputFile = args[1];
// Create a Java Spark Context.
SparkConf conf = new SparkConf().setMaster("mesos://10.230.136.197:5050").setAppName("wordCount").set("spark.executor.uri", "http://d3kbcqa49mib13.cloudfront.net/spark-1.5.1-bin-hadoop2.6.tgz");
JavaSparkContext sc = new JavaSparkContext(conf);
// Load our input data.
JavaRDD<String> input = sc.textFile(inputFile);
// Split up into words.
JavaRDD<String> words = input.flatMap(
new FlatMapFunction<String, String>() {
public Iterable<String> call(String x) {
return Arrays.asList(x.split(" "));
}});
// Transform into word and count.
JavaPairRDD<String, Integer> counts = words.mapToPair(
new PairFunction<String, String, Integer>(){
public Tuple2<String, Integer> call(String x){
return new Tuple2(x, 1);
}}).reduceByKey(new Function2<Integer, Integer, Integer>(){
public Integer call(Integer x, Integer y){ return x + y;}});
// Save the word count back out to a text file, causing evaluation.
counts.saveAsTextFile(outputFile);
}
}
进入到examples/mini-complete-example 中，重新编译生成jar包。
4）启动spark-mesos-dispather
./sbin/start-mesos-dispatcher.sh -m mesos://10.230.136.197:5050
Spark Command: /app/otter/jdk1..0_80/bin/java -cp /home/qingpingzhang/dev/spark-1.5.-bin-hadoop2./sbin/../conf/:/home/qingpingzhang/dev/spark-1.5.-bin-hadoop2./lib/spark-assembly-1.5.-hadoop2.6.0.jar:/home/qingpingzhang/dev/spark-1.5.-bin-hadoop2./lib/datanucleus-core-3.2..jar:/home/qingpingzhang/dev/spark-1.5.-bin-hadoop2./lib/datanucleus-rdbms-3.2..jar:/home/qingpingzhang/dev/spark-1.5.-bin-hadoop2./lib/datanucleus-api-jdo-3.2..jar -Xms1g -Xmx1g -XX:MaxPermSize=256m org.apache.spark.deploy.mesos.MesosClusterDispatcher --host vg-log-analysis-prod --port -m mesos://10.230.136.197:5050
========================================
Using Spark's default log4j profile: org/apache/spark/log4j-defaults.properties
// :: INFO MesosClusterDispatcher: Registered signal handlers for [TERM, HUP, INT]
// :: WARN Utils: Your hostname, vg-log-analysis-prod resolves to a loopback address: 127.0.0.1; using 10.230.136.197 instead (on interface eth0)
// :: WARN Utils: Set SPARK_LOCAL_IP if you need to bind to another address
// :: INFO MesosClusterDispatcher: Recovery mode in Mesos dispatcher set to: NONE
// :: WARN NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
// :: INFO SecurityManager: Changing view acls to: qingpingzhang
// :: INFO SecurityManager: Changing modify acls to: qingpingzhang
// :: INFO SecurityManager: SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(qingpingzhang); users with modify permissions: Set(qingpingzhang)
// :: INFO SecurityManager: Changing view acls to: qingpingzhang
// :: INFO SecurityManager: Changing modify acls to: qingpingzhang
// :: INFO SecurityManager: SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(qingpingzhang); users with modify permissions: Set(qingpingzhang)
// :: INFO Utils: Successfully started service on port .
// :: INFO MesosClusterUI: Started MesosClusterUI at http://10.230.136.197:8081
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1105 ::37.927594 sched.cpp:]
**************************************************
Scheduler driver bound to loopback interface! Cannot communicate with remote master(s). You might want to set 'LIBPROCESS_IP' environment variable to use a routable IP address.
**************************************************
I1105 03:31:37.931098 3408 sched.cpp:164] Version: 0.24.0
I1105 03:31:37.939507 3406 sched.cpp:262] New master detected at master@10.230.136.197:5050
I1105 03:31:37.940353 3406 sched.cpp:272] No credentials provided. Attempting to register without authentication
I1105 03:31:37.943528 3406 sched.cpp:640] Framework registered with 20151105-021937-16777343-5050-32543-0001
15/11/05 11:31:37 INFO MesosClusterScheduler: Registered as framework ID 20151105-021937-16777343-5050-32543-0001
// :: INFO Utils: Successfully started service on port .
// :: INFO MesosRestServer: Started REST server for submitting applications on port
5)提交任务
./bin/spark-submit --master mesos://10.230.136.197:7077 --deploy-mode cluster --class com.oreilly.learningsparkexamples.mini.java.WordCount /home/qingpingzhang/dev/spark-1.5.1-bin-hadoop2.6/examples/mini-complete-example/target/learning-spark-mini-example-0.0.1.jar /home/qingpingzhang/dev/spark-1.5.1-bin-hadoop2.6/README.md /home/qingpingzhang/dev/spark-1.5.1-bin-hadoop2.6/wordcounts.txt
然后我们可以看到mesos-master的输出，收到任务，然后发送给slaver，最后更新任务状态：
I1105 05:08:33.312283 7490 master.cpp:2094] Received SUBSCRIBE call for framework 'Spark Cluster' at scheduler-fae58488-7d56-4661-a078-938d12871930@10.230.136.197:57392
I1105 ::33.312641 master.cpp:] Subscribing framework Spark Cluster with checkpointing enabled and capabilities [ ]
I1105 ::33.313761 hierarchical.hpp:] Added framework -----
I1105 05:08:33.315335 7490 master.cpp:4613] Sending 1 offers to framework 20151105-050710-3314083338-5050-7469-0000 (Spark Cluster) at scheduler-fae58488-7d56-4661-a078-938d12871930@10.230.136.197:57392
I1105 ::33.426009 master.cpp:] Processing ACCEPT call for offers: [ -----O0 ] on slave -----S0 at slave()@10.29.23.28: (ip----.ec2.internal) for framework --$
--- (Spark Cluster) at scheduler-fae58488-7d56--a078-938d12871930@10.230.136.197:
I1105 ::33.427104 hierarchical.hpp:] Recovered cpus(*):; mem(*):; disk(*):; ports(*):[-] (total: cpus(*):; mem(*):; disk(*):; ports(*):[-], allocated: ) on slave -----S0 from framework
-----
I1105 ::39.177790 master.cpp:] Sending offers to framework ----- (Spark Cluster) at scheduler-fae58488-7d56--a078-938d12871930@10.230.136.197:
I1105 ::39.181149 master.cpp:] Processing ACCEPT call for offers: [ -----O1 ] on slave -----S0 at slave()@10.29.23.28: (ip----.ec2.internal) for framework --$
--- (Spark Cluster) at scheduler-fae58488-7d56--a078-938d12871930@10.230.136.197:
I1105 ::39.181699 hierarchical.hpp:] Recovered cpus(*):; mem(*):; disk(*):; ports(*):[-] (total: cpus(*):; mem(*):; disk(*):; ports(*):[-], allocated: ) on slave -----S0 from framework
-----
I1105 ::44.183100 master.cpp:] Sending offers to framework ----- (Spark Cluster) at scheduler-fae58488-7d56--a078-938d12871930@10.230.136.197:
I1105 ::44.186468 master.cpp:] Processing ACCEPT call for offers: [ -----O2 ] on slave -----S0 at slave()@10.29.23.28: (ip----.ec2.internal) for framework --$
--- (Spark Cluster) at scheduler-fae58488-7d56--a078-938d12871930@10.230.136.197:
I1105 ::44.187100 hierarchical.hpp:] Recovered cpus(*):; mem(*):; disk(*):; ports(*):[-] (total: cpus(*):; mem(*):; disk(*):; ports(*):[-], allocated: ) on slave -----S0 from framework
----- I1105 ::18.668609 master.cpp:] Status update TASK_FAILED (UUID: 8d30c637-b885-487b-b174-47232cc0e49f) for task driver-- of framework ----- from slave -----S0 at slave()@
.29.23.: (ip----.ec2.internal)
I1105 ::18.668689 master.cpp:] Forwarding status update TASK_FAILED (UUID: 8d30c637-b885-487b-b174-47232cc0e49f) for task driver-- of framework -----
I1105 05:12:18.669001 7489 master.cpp:5576] Updating the latest state of task driver-20151105131213-0002 of framework 20151105-050710-3314083338-5050-7469-0000 to TASK_FAILED
I1105 ::18.669373 hierarchical.hpp:] Recovered cpus(*):; mem(*): (total: cpus(*):; mem(*):; disk(*):; ports(*):[-], allocated: ) on slave -----S0 from framework ----- I1105 ::18.670912 master.cpp:] Removing task driver-- with resources cpus(*):; mem(*): of framework ----- on slave -----S0 at slave()@10.29.23.28: (ip----
.ec2.internal)
mesos-slaver的输出，收到任务，然后下载spark的可执行文件失败,最后执行任务失败：
1105 05:11:31.363765 17084 slave.cpp:1270] Got assigned task driver-20151105131130-0001 for framework 20151105-050710-3314083338-5050-7469-0000
I1105 05:11:31.365025 17084 slave.cpp:1386] Launching task driver-20151105131130-0001 for framework 20151105-050710-3314083338-5050-7469-0000
I1105 05:11:31.376075 17084 slave.cpp:4852] Launching executor driver-20151105131130-0001 of framework 20151105-050710-3314083338-5050-7469-0000 with resources cpus(*):0.1; mem(*):32 in work directory '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/$
0151105-050710-3314083338-5050-7469-0000/executors/driver-20151105131130-0001/runs/461ceb14-9247-4a59-b9d4-4b0a7947e353'
I1105 05:11:31.376448 17085 containerizer.cpp:640] Starting container '461ceb14-9247-4a59-b9d4-4b0a7947e353' for executor 'driver-20151105131130-0001' of framework '20151105-050710-3314083338-5050-7469-0000'
I1105 05:11:31.376878 17084 slave.cpp:1604] Queuing task 'driver-20151105131130-0001' for executor driver-20151105131130-0001 of framework '20151105-050710-3314083338-5050-7469-0000
I1105 ::31.379096 linux_launcher.cpp:] Cloning child process with flags =
I1105 ::31.382968 containerizer.cpp:] Checkpointing executor's forked pid 17098 to '/tmp/mesos/meta/slaves/-----S0/frameworks/-----/executors/driver--/runs/461ceb14--4a$
-b9d4-4b0a7947e353/pids/forked.pid'
E1105 05:11:31.483093 17078 fetcher.cpp:515] Failed to run mesos-fetcher: Failed to fetch all URIs for container '461ceb14-9247-4a59-b9d4-4b0a7947e353' with exit status: 256
E1105 ::31.483355 slave.cpp:] Container '461ceb14-9247-4a59-b9d4-4b0a7947e353' for executor 'driver-20151105131130-0001' of framework '20151105-050710-3314083338-5050-7469-0000' failed to start: Failed to fetch all URIs for container '461ceb14-9247-4a59-b9d$
-4b0a7947e353' with exit status: 256
I1105 ::31.483444 containerizer.cpp:] Destroying container '461ceb14-9247-4a59-b9d4-4b0a7947e353'
I1105 ::31.485548 cgroups.cpp:] Freezing cgroup /sys/fs/cgroup/freezer/mesos/461ceb14--4a59-b9d4-4b0a7947e353
I1105 ::31.487112 cgroups.cpp:] Successfully froze cgroup /sys/fs/cgroup/freezer/mesos/461ceb14--4a59-b9d4-4b0a7947e353 after .48992ms
I1105 ::31.488673 cgroups.cpp:] Thawing cgroup /sys/fs/cgroup/freezer/mesos/461ceb14--4a59-b9d4-4b0a7947e353
I1105 ::31.490102 cgroups.cpp:] Successfullly thawed cgroup /sys/fs/cgroup/freezer/mesos/461ceb14--4a59-b9d4-4b0a7947e353 after .363968ms
I1105 ::31.583328 containerizer.cpp:] Executor for container '461ceb14-9247-4a59-b9d4-4b0a7947e353' has exited
I1105 ::31.583977 slave.cpp:] Executor 'driver-20151105131130-0001' of framework ----- exited with status
I1105 ::31.585134 slave.cpp:] Handling status update TASK_FAILED (UUID: 0c429164-a159-4e72--c5fbd2ef9004) for task driver-- of framework ----- from @0.0.0.0:
W1105 ::31.585384 containerizer.cpp:] Ignoring update for unknown container: 461ceb14--4a59-b9d4-4b0a7947e353
I1105 ::31.585605 status_update_manager.cpp:] Received status update TASK_FAILED (UUID: 0c429164-a159-4e72--c5fbd2ef9004) for task driver-- of framework -----
I1105 ::31.585911 status_update_manager.cpp:] Checkpointing UPDATE for status update TASK_FAILED (UUID: 0c429164-a159-4e72--c5fbd2ef9004) for task driver-- of framework -----
I1105 05:11:31.596305 17081 slave.cpp:3016] Forwarding the update TASK_FAILED (UUID: 0c429164-a159-4e72-9872-c5fbd2ef9004) for task driver-20151105131130-0001 of framework 20151105-050710-3314083338-5050-7469-0000 to master@10.230.136.197:5050
I1105 ::31.611620 status_update_manager.cpp:] Received status update acknowledgement (UUID: 0c429164-a159-4e72--c5fbd2ef9004) for task driver-- of framework -----
I1105 ::31.611702 status_update_manager.cpp:] Checkpointing ACK for status update TASK_FAILED (UUID: 0c429164-a159-4e72--c5fbd2ef9004) for task driver-- of framework -----
I1105 ::31.616345 slave.cpp:] Cleaning up executor 'driver-20151105131130-0001' of framework -----
那么这个问题咋解决呢？
去mesos的slaver上面查看运行日志（默认是在/tmp/mesos/slaves/目录下），发现尼玛，这个跟stand alone还不一样（stand alone模式下面，spark的master会建立一个http server，把jar包提供给spark worker下载执行，但是mesos模式居然不是这样，jar包需要是放到hdfs 或者http 路径下面？坑爹了），这有点不合理啊，亲。
/tmp/mesos/slaves/-----S0/frameworks/-----/executors/driver--/runs/bed1c620-e849--b130-c95c47133599$ cat stderr
I1105 ::17.387609 fetcher.cpp:] Fetcher Info: {"cache_directory":"\/tmp\/mesos\/fetch\/slaves\/20151105-045733-3314083338-5050-7152-S0\/qingpingzhang","items":[{"action":"BYPASS_CACHE","uri":{"extract":true,"value":"\/home\/qingpingzhang\/dev\/spark-1.5.1-bin-hadoop2.6\/examples\/mini-complete-example\/target\/learning-spark-mini-example-0.0.1.jar"}},{"action":"BYPASS_CACHE","uri":{"extract":true,"value":"http:\/\/d3kbcqa49mib13.cloudfront.net\/spark-1.5.1-bin-hadoop2.6.tgz"}}],"sandbox_directory":"\/tmp\/mesos\/slaves\/20151105-045733-3314083338-5050-7152-S0\/frameworks\/20151105-070418-3314083338-5050-12075-0000\/executors\/driver-20151105151012-0001\/runs\/bed1c620-e849-4076-b130-c95c47133599","user":"qingpingzhang"}
I1105 07:10:17.390316 17897 fetcher.cpp:369] Fetching URI '/home/qingpingzhang/dev/spark-1.5.1-bin-hadoop2.6/examples/mini-complete-example/target/learning-spark-mini-example-0.0.1.jar'
I1105 07:10:17.390344 17897 fetcher.cpp:243] Fetching directly into the sandbox directory
I1105 07:10:17.390384 17897 fetcher.cpp:180] Fetching URI '/home/qingpingzhang/dev/spark-1.5.1-bin-hadoop2.6/examples/mini-complete-example/target/learning-spark-mini-example-0.0.1.jar'
I1105 07:10:17.390418 17897 fetcher.cpp:160] Copying resource with command:cp '/home/qingpingzhang/dev/spark-1.5.1-bin-hadoop2.6/examples/mini-complete-example/target/learning-spark-mini-example-0.0.1.jar' '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000/executors/driver-20151105151012-0001/runs/bed1c620-e849-4076-b130-c95c47133599/learning-spark-mini-example-0.0.1.jar'
cp: cannot stat ‘/home/qingpingzhang/dev/spark-1.5.-bin-hadoop2./examples/mini-complete-example/target/learning-spark-mini-example-0.0..jar’: No such file or directory
Failed to fetch '/home/qingpingzhang/dev/spark-1.5.1-bin-hadoop2.6/examples/mini-complete-example/target/learning-spark-mini-example-0.0.1.jar': Failed to copy with command 'cp '/home/qingpingzhang/dev/spark-1.5.-bin-hadoop2./examples/mini-complete-example/target/learning-spark-mini-example-0.0..jar' '/tmp/mesos/slaves/-----S0/frameworks/-----/executors/driver--/runs/bed1c620-e849--b130-c95c47133599/learning-spark-mini-example-0.0..jar'', exit status:
Failed to synchronize with slave (it's probably exited)
好吧，为了能够跑通测试，现在把jar包和readme.txt文件copy到mesos slaver的机器上面去。
./bin/spark-submit --master mesos://10.230.136.197:7077 --deploy-mode cluster --class com.oreilly.learningsparkexamples.mini.java.WordCount /tmp/learning-spark-mini-example-0.0.1.jar /tmp/README.md /tmp/wordcounts.txt
果然任务就执行成功了,meso-slave输入出如下：
I1105 ::26.748515 slave.cpp:] Got assigned task driver-- for framework -----
I1105 ::26.749575 gc.cpp:] Unscheduling '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000' from gc
I1105 ::26.749703 gc.cpp:] Unscheduling '/tmp/mesos/meta/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000' from gc
I1105 ::26.749825 slave.cpp:] Launching task driver-- for framework -----
I1105 ::26.760673 slave.cpp:] Launching executor driver-- of framework ----- with resources cpus(*):0.1; mem(*): in work directory '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks$
-----/executors/driver--/runs/910a4983--41dd-b014-66827c044c16'
I1105 ::26.760967 containerizer.cpp:] Starting container '910a4983-2732-41dd-b014-66827c044c16' for executor 'driver-20151105154326-0002' of framework '20151105-070418-3314083338-5050-12075-0000'
I1105 ::26.761265 slave.cpp:] Queuing task 'driver-20151105154326-0002' for executor driver-- of framework '20151105-070418-3314083338-5050-12075-0000
I1105 ::26.763134 linux_launcher.cpp:] Cloning child process with flags =
I1105 ::26.766726 containerizer.cpp:] Checkpointing executor's forked pid 18129 to '/tmp/mesos/meta/slaves/-----S0/frameworks/-----/executors/driver--/runs/910a4983--$dd-b014-66827c044c16/pids/forked.pid'
I1105 ::33.153153 slave.cpp:] Got registration for executor 'driver-20151105154326-0002' of framework ----- from executor()@10.29.23.28:
I1105 07:43:33.154284 17246 slave.cpp:1760] Sending queued task 'driver-20151105154326-0002' to executor 'driver-20151105154326-0002' of framework 20151105-070418-3314083338-5050-12075-0000
I1105 07:43:33.160464 17243 slave.cpp:2717] Handling status update TASK_RUNNING (UUID: eb83c08d-19fc-4d20-ab17-6f69ee8d90b2) for task driver-20151105154326-0002 of framework 20151105-070418-3314083338-5050-12075-0000 from executor(1)@10.29.23.28:54580
I1105 07:43:33.160643 17242 status_update_manager.cpp:322] Received status update TASK_RUNNING (UUID: eb83c08d-19fc-4d20-ab17-6f69ee8d90b2) for task driver-20151105154326-0002 of framework 20151105-070418-3314083338-5050-12075-0000
I1105 ::33.160940 status_update_manager.cpp:] Checkpointing UPDATE for status update TASK_RUNNING (UUID: eb83c08d-19fc-4d20-ab17-6f69ee8d90b2) for task driver-- of framework -----
I1105 ::33.168092 slave.cpp:] Forwarding the update TASK_RUNNING (UUID: eb83c08d-19fc-4d20-ab17-6f69ee8d90b2) for task driver-- of framework ----- to master@10.230.136.197:
I1105 ::33.168218 slave.cpp:] Sending acknowledgement for status update TASK_RUNNING (UUID: eb83c08d-19fc-4d20-ab17-6f69ee8d90b2) for task driver-- of framework ----- to executor()@10.29.23.28:
I1105 ::33.171906 status_update_manager.cpp:] Received status update acknowledgement (UUID: eb83c08d-19fc-4d20-ab17-6f69ee8d90b2) for task driver-- of framework -----
I1105 ::33.172025 status_update_manager.cpp:] Checkpointing ACK for status update TASK_RUNNING (UUID: eb83c08d-19fc-4d20-ab17-6f69ee8d90b2) for task driver-- of framework -----
I1105 ::36.454344 slave.cpp:] Current disk usage 1.10%. Max allowed age: .223128215149259days
I1105 ::39.174698 slave.cpp:] Got assigned task for framework -----
I1105 ::39.175014 slave.cpp:] Launching task for framework -----
I1105 ::39.185343 slave.cpp:] Launching executor -----S0 of framework ----- with resources cpus(*):; mem(*): in work directory '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-$
/frameworks/-----/executors/-----S0/runs/8b233e11-54a6-41b6-a8ad-f82660d640b5'
I1105 ::39.185643 slave.cpp:] Queuing task '' for executor -----S0 of framework '20151105-070418-3314083338-5050-12075-0001
I1105 ::39.185694 containerizer.cpp:] Starting container '8b233e11-54a6-41b6-a8ad-f82660d640b5' for executor '20151105-045733-3314083338-5050-7152-S0' of framework '20151105-070418-3314083338-5050-12075-0001'
I1105 ::39.185786 slave.cpp:] Got assigned task for framework -----
I1105 ::39.185931 slave.cpp:] Launching task for framework -----
I1105 ::39.185968 slave.cpp:] Queuing task '' for executor -----S0 of framework '20151105-070418-3314083338-5050-12075-0001
I1105 ::39.187925 linux_launcher.cpp:] Cloning child process with flags =
I1105 ::46.388809 slave.cpp:] Got registration for executor '20151105-045733-3314083338-5050-7152-S0' of framework ----- from executor()@10.29.23.28:
I1105 ::46.389571 slave.cpp:] Sending queued task '' to executor '20151105-045733-3314083338-5050-7152-S0' of framework -----
I1105 ::46.389883 slave.cpp:] Sending queued task '' to executor '20151105-045733-3314083338-5050-7152-S0' of framework -----
I1105 ::49.534858 slave.cpp:] Handling status update TASK_RUNNING (UUID: b6d77a8f-ef6f-414b-9bdc-0fa1c2a96841) for task of framework ----- from executor()@10.29.23.28:
I1105 ::49.535087 slave.cpp:] Handling status update TASK_RUNNING (UUID: 6633c11b-c403-45c6-82c5-f495fcc9ae70) for task of framework ----- from executor()@10.29.23.28:
#......更多日志....
I1105 ::53.012852 slave.cpp:] Sending acknowledgement for status update TASK_FINISHED (UUID: 72c9805d-ea9d-49c1-ac2a-5c3940aa77f5) for task of framework ----- to executor()@10.29.23.28:
I1105 ::53.016926 status_update_manager.cpp:] Received status update acknowledgement (UUID: 5f175f09-2a35-46dd-9ac9-fbe648d9780a) for task of framework -----
I1105 ::53.017357 status_update_manager.cpp:] Received status update acknowledgement (UUID: 72c9805d-ea9d-49c1-ac2a-5c3940aa77f5) for task of framework -----
I1105 ::53.146410 slave.cpp:] Asked to shut down framework ----- by master@10.230.136.197:
I1105 ::53.146461 slave.cpp:] Shutting down framework -----
I1105 ::53.146515 slave.cpp:] Shutting down executor '20151105-045733-3314083338-5050-7152-S0' of framework -----
I1105 ::53.637172 slave.cpp:] Handling status update TASK_FINISHED (UUID: 9fad65ef-d651-4d2c-91a7-bac8cf4b55a4) for task driver-- of framework ----- from executor()@10.29.23.28:
I1105 ::53.637706 status_update_manager.cpp:] Received status update TASK_FINISHED (UUID: 9fad65ef-d651-4d2c-91a7-bac8cf4b55a4) for task driver-- of framework -----
I1105 07:43:53.637755 17246 status_update_manager.cpp:826] Checkpointing UPDATE for status update TASK_FINISHED (UUID: 9fad65ef-d651-4d2c-91a7-bac8cf4b55a4) for task driver-20151105154326-0002 of framework 20151105-070418-3314083338-5050-12075-0000
I1105 07:43:53.643517 17242 slave.cpp:3016] Forwarding the update TASK_FINISHED (UUID: 9fad65ef-d651-4d2c-91a7-bac8cf4b55a4) for task driver-20151105154326-0002 of framework 20151105-070418-3314083338-5050-12075-0000 to master@10.230.136.197:5050
I1105 ::53.643635 slave.cpp:] Sending acknowledgement for status update TASK_FINISHED (UUID: 9fad65ef-d651-4d2c-91a7-bac8cf4b55a4) for task driver-- of framework ----- to executor()@10.29.23.28:
I1105 ::53.647647 status_update_manager.cpp:] Received status update acknowledgement (UUID: 9fad65ef-d651-4d2c-91a7-bac8cf4b55a4) for task driver-- of framework -----
I1105 ::53.647703 status_update_manager.cpp:] Checkpointing ACK for status update TASK_FINISHED (UUID: 9fad65ef-d651-4d2c-91a7-bac8cf4b55a4) for task driver-- of framework -----
I1105 ::54.689678 containerizer.cpp:] Executor for container '910a4983-2732-41dd-b014-66827c044c16' has exited
I1105 ::54.689708 containerizer.cpp:] Destroying container '910a4983-2732-41dd-b014-66827c044c16'
I1105 ::54.691368 cgroups.cpp:] Freezing cgroup /sys/fs/cgroup/freezer/mesos/910a4983--41dd-b014-66827c044c16
I1105 ::54.693023 cgroups.cpp:] Successfully froze cgroup /sys/fs/cgroup/freezer/mesos/910a4983--41dd-b014-66827c044c16 after .624064ms
I1105 ::54.694628 cgroups.cpp:] Thawing cgroup /sys/fs/cgroup/freezer/mesos/910a4983--41dd-b014-66827c044c16
I1105 ::54.695976 cgroups.cpp:] Successfullly thawed cgroup /sys/fs/cgroup/freezer/mesos/910a4983--41dd-b014-66827c044c16 after 1312us
I1105 ::54.697335 slave.cpp:] Executor 'driver-20151105154326-0002' of framework ----- exited with status
I1105 ::54.697371 slave.cpp:] Cleaning up executor 'driver-20151105154326-0002' of framework -----
I1105 ::54.697621 gc.cpp:] Scheduling '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000/executors/driver-20151105154326-0002/runs/910a4983-2732-41dd-b014-66827c044c16' for gc .99999192621333days i
n the future
I1105 ::54.697669 gc.cpp:] Scheduling '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000/executors/driver-20151105154326-0002' for gc .99999192552296days in the future
I1105 ::54.697700 gc.cpp:] Scheduling '/tmp/mesos/meta/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000/executors/driver-20151105154326-0002/runs/910a4983-2732-41dd-b014-66827c044c16' for gc 6.99999192509333d
ays in the future
I1105 ::54.697713 slave.cpp:] Cleaning up framework -----
I1105 ::54.697726 gc.cpp:] Scheduling '/tmp/mesos/meta/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000/executors/driver-20151105154326-0002' for gc .99999192474667days in the future
I1105 ::54.697756 status_update_manager.cpp:] Closing status update streams for framework -----
I1105 ::54.697813 gc.cpp:] Scheduling '/tmp/mesos/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000' for gc .99999192391407days in the future
I1105 ::54.697856 gc.cpp:] Scheduling '/tmp/mesos/meta/slaves/20151105-045733-3314083338-5050-7152-S0/frameworks/20151105-070418-3314083338-5050-12075-0000' for gc .99999192340148days in the future
I1105 ::58.147456 slave.cpp:] Killing executor '20151105-045733-3314083338-5050-7152-S0' of framework -----
I1105 ::58.147546 containerizer.cpp:] Destroying container '8b233e11-54a6-41b6-a8ad-f82660d640b5'
I1105 ::58.149194 cgroups.cpp:] Freezing cgroup /sys/fs/cgroup/freezer/mesos/8b233e11-54a6-41b6-a8ad-f82660d640b5
I1105 ::58.200407 containerizer.cpp:] Executor for container '8b233e11-54a6-41b6-a8ad-f82660d640b5' has exited
然后/tmp/wordcounts.txt/目录下也如愿以偿的出现了统计结果：
ll /tmp/wordcounts.txt/
total
drwxr-xr-x qingpingzhang qingpingzhang Nov : ./
drwxrwxrwt root root Nov : ../
-rw-r--r-- qingpingzhang qingpingzhang Nov : part-
-rw-r--r-- qingpingzhang qingpingzhang Nov : .part-.crc
-rw-r--r-- qingpingzhang qingpingzhang Nov : part-
-rw-r--r-- qingpingzhang qingpingzhang Nov : .part-.crc
-rw-r--r-- qingpingzhang qingpingzhang Nov : _SUCCESS
-rw-r--r-- qingpingzhang qingpingzhang Nov : ._SUCCESS.crc
至此，spark 运行在mesos框架上也算是跑通了。
这里没有把mesos，spark的webui 中的图贴出来。
总结一下：
1）搭建好mesos集群
2）安装好spark，启动spark的mesos-dispather（连接到mesos的matser上）
3）采用spark-submit来提交任务（提交任务的时候，需要把jar包放到可以下载的地方，例如：hdfs，http等）
-----------------
使用mesos的好处，主要是能够动态的启动和分配任务，尽量利用好机器资源。
由于我们的日志分析就2台机器，所以还是打算用spark的stand alone 模式启动。