伪分布式安装部署CDH4.2.1与Impala[原创实践]


参考资料:
http://www.cloudera.com/content/cloudera-content/cloudera-docs/CDH4/latest/CDH4-Quick-Start/cdh4qs_topic_3_3.html
http://www.cloudera.com/content/cloudera-content/cloudera-docs/Impala/latest/Installing-and-Using-Impala/Installing-and-Using-Impala.html
http://blog.cloudera.com/blog/2013/02/from-zero-to-impala-in-minutes/

什么是Impala?
Cloudera发布了实时查询开源项目Impala,根据多款产品实测表明,它比原来基于MapReduce的Hive SQL查询速度提升3~90倍。Impala是Google Dremel的模仿,但在SQL功能上青出于蓝胜于蓝。

1. 安装JDK
$ sudo yum install jdk-6u41-linux-amd64.rpm

2. 伪分布式模式安装CDH4
$ cd /etc/yum.repos.d/
$ sudo wget http://archive.cloudera.com/cdh4/redhat/6/x86_64/cdh/cloudera-cdh4.repo
$ sudo yum install hadoop-conf-pseudo

格式化NameNode.
$ sudo -u hdfs hdfs namenode -format

启动HDFS
$ for x in `cd /etc/init.d ; ls hadoop-hdfs-*` ; do sudo service $x start ; done

创建/tmp目录
$ sudo -u hdfs hadoop fs -rm -r /tmp
$ sudo -u hdfs hadoop fs -mkdir /tmp
$ sudo -u hdfs hadoop fs -chmod -R 1777 /tmp

创建YARN与日志目录
$ sudo -u hdfs hadoop fs -mkdir /tmp/hadoop-yarn/staging
$ sudo -u hdfs hadoop fs -chmod -R 1777 /tmp/hadoop-yarn/staging

$ sudo -u hdfs hadoop fs -mkdir /tmp/hadoop-yarn/staging/history/done_intermediate
$ sudo -u hdfs hadoop fs -chmod -R 1777 /tmp/hadoop-yarn/staging/history/done_intermediate

$ sudo -u hdfs hadoop fs -chown -R mapred:mapred /tmp/hadoop-yarn/staging

$ sudo -u hdfs hadoop fs -mkdir /var/log/hadoop-yarn
$ sudo -u hdfs hadoop fs -chown yarn:mapred /var/log/hadoop-yarn

检查HDFS文件树
$ sudo -u hdfs hadoop fs -ls -R /

drwxrwxrwt - hdfs supergroup 0 2012-05-31 15:31 /tmp
drwxr-xr-x - hdfs supergroup 0 2012-05-31 15:31 /tmp/hadoop-yarn
drwxrwxrwt - mapred mapred 0 2012-05-31 15:31 /tmp/hadoop-yarn/staging
drwxr-xr-x - mapred mapred 0 2012-05-31 15:31 /tmp/hadoop-yarn/staging/history
drwxrwxrwt - mapred mapred 0 2012-05-31 15:31 /tmp/hadoop-yarn/staging/history/done_intermediate
drwxr-xr-x - hdfs supergroup 0 2012-05-31 15:31 /var
drwxr-xr-x - hdfs supergroup 0 2012-05-31 15:31 /var/log
drwxr-xr-x - yarn mapred 0 2012-05-31 15:31 /var/log/hadoop-yarn

启动YARN
$ sudo service hadoop-yarn-resourcemanager start
$ sudo service hadoop-yarn-nodemanager start
$ sudo service hadoop-mapreduce-historyserver start

创建用户目录(以用户dong.guo为例):
$ sudo -u hdfs hadoop fs -mkdir /user/dong.guo
$ sudo -u hdfs hadoop fs -chown dong.guo /user/dong.guo

测试上传文件
$ hadoop fs -mkdir input
$ hadoop fs -put /etc/hadoop/conf/*.xml input
$ hadoop fs -ls input

Found 4 items
-rw-r--r--   1 dong.guo supergroup       1461 2013-05-14 03:30 input/core-site.xml
-rw-r--r--   1 dong.guo supergroup       1854 2013-05-14 03:30 input/hdfs-site.xml
-rw-r--r--   1 dong.guo supergroup       1325 2013-05-14 03:30 input/mapred-site.xml
-rw-r--r--   1 dong.guo supergroup       2262 2013-05-14 03:30 input/yarn-site.xml

配置HADOOP_MAPRED_HOME环境变量
$ export HADOOP_MAPRED_HOME=/usr/lib/hadoop-mapreduce

运行一个测试Job
$ hadoop jar /usr/lib/hadoop-mapreduce/hadoop-mapreduce-examples.jar grep input output23 'dfs[a-z.]+'

Job完成后,可以看到以下目录
$ hadoop fs -ls

Found 2 items
drwxr-xr-x   - dong.guo supergroup          0 2013-05-14 03:30 input
drwxr-xr-x   - dong.guo supergroup          0 2013-05-14 03:32 output23

$ hadoop fs -ls output23

Found 2 items
-rw-r--r--   1 dong.guo supergroup          0 2013-05-14 03:32 output23/_SUCCESS
-rw-r--r--   1 dong.guo supergroup        150 2013-05-14 03:32 output23/part-r-00000

$ hadoop fs -cat output23/part-r-00000 | head

1	dfs.safemode.min.datanodes
1	dfs.safemode.extension
1	dfs.replication
1	dfs.namenode.name.dir
1	dfs.namenode.checkpoint.dir
1	dfs.datanode.data.dir

3. 安装 Hive
$ sudo yum install hive hive-metastore hive-server

$ sudo yum install mysql-server

$ sudo service mysqld start

$ cd ~
$ wget 'http://cdn.mysql.com/Downloads/Connector-J/mysql-connector-java-5.1.25.tar.gz'
$ tar xzf mysql-connector-java-5.1.25.tar.gz
$ sudo cp mysql-connector-java-5.1.25/mysql-connector-java-5.1.25-bin.jar /usr/lib/hive/lib/

$ sudo /usr/bin/mysql_secure_installation

[...]
Enter current password for root (enter for none):
OK, successfully used password, moving on...
[...]
Set root password? [Y/n] y
New password:hadoophive
Re-enter new password:hadoophive
Remove anonymous users? [Y/n] Y
[...]
Disallow root login remotely? [Y/n] N
[...]
Remove test database and access to it [Y/n] Y
[...]
Reload privilege tables now? [Y/n] Y
All done!

$ mysql -u root -phadoophive

mysql> CREATE DATABASE metastore;
mysql> USE metastore;
mysql> SOURCE /usr/lib/hive/scripts/metastore/upgrade/mysql/hive-schema-0.10.0.mysql.sql;

mysql> CREATE USER 'hive'@'%' IDENTIFIED BY 'hadoophive';
mysql> CREATE USER 'hive'@'localhost' IDENTIFIED BY 'hadoophive';
mysql> REVOKE ALL PRIVILEGES, GRANT OPTION FROM 'hive'@'%';
mysql> REVOKE ALL PRIVILEGES, GRANT OPTION FROM 'hive'@'localhost';
mysql> GRANT SELECT,INSERT,UPDATE,DELETE,LOCK TABLES,EXECUTE ON metastore.* TO 'hive'@'%';
mysql> GRANT SELECT,INSERT,UPDATE,DELETE,LOCK TABLES,EXECUTE ON metastore.* TO 'hive'@'localhost';
mysql> FLUSH PRIVILEGES;
mysql> quit;

$ sudo mv /etc/hive/conf/hive-site.xml /etc/hive/conf/hive-site.xml.bak
$ sudo vim /etc/hive/conf/hive-site.xml






  javax.jdo.option.ConnectionURL
  jdbc:mysql://localhost/metastore
  the URL of the MySQL database



  javax.jdo.option.ConnectionDriverName
  com.mysql.jdbc.Driver



  javax.jdo.option.ConnectionUserName
  hive



  javax.jdo.option.ConnectionPassword
  hadoophive



  datanucleus.autoCreateSchema
  false



  datanucleus.fixedDatastore
  true



  hive.metastore.uris
  thrift://127.0.0.1:9083
  IP address (or fully-qualified domain name) and port of the metastore host



  hive.aux.jars.path
  file:///usr/lib/hive/lib/zookeeper.jar,file:///usr/lib/hive/lib/hbase.jar,file:///usr/lib/hive/lib/hive-hbase-handler-0.10.0-cdh4.2.0.jar,file:///usr/lib/hive/lib/guava-11.0.2.jar



$ sudo service hive-metastore start

Starting (hive-metastore):                                 [  OK  ]

$ sudo service hive-server start

Starting (hive-server):                                    [  OK  ]

$ sudo -u hdfs hadoop fs -mkdir /user/hive
$ sudo -u hdfs hadoop fs -chown hive /user/hive
$ sudo -u hdfs hadoop fs -mkdir /tmp
$ sudo -u hdfs hadoop fs -chmod 777 /tmp
$ sudo -u hdfs hadoop fs -chmod o+t /tmp
$ sudo -u hdfs hadoop fs -mkdir /data
$ sudo -u hdfs hadoop fs -chown hdfs /data
$ sudo -u hdfs hadoop fs -chmod 777 /data
$ sudo -u hdfs hadoop fs -chmod o+t /data

$ sudo chown -R hive:hive /var/lib/hive
$ sudo vim /tmp/kv1.txt

1	www.baidu.com
2	www.google.com
3	www.sina.com.cn
4	www.163.com
5	heylinx.com

$ sudo -u hive hive

Logging initialized using configuration in file:/etc/hive/conf.dist/hive-log4j.properties
Hive history file=/tmp/root/hive_job_log_root_201305140801_825709760.txt

hive> CREATE TABLE IF NOT EXISTS pokes ( foo INT,bar STRING ) ROW FORMAT DELIMITED FIELDS TERMINATED BY "\t" LINES TERMINATED BY "\n";

hive> show tables;
OK
pokes
Time taken: 0.415 seconds
hive> LOAD DATA LOCAL INPATH '/tmp/kv1.txt' OVERWRITE INTO TABLE pokes;
Copying data from file:/tmp/kv1.txt
Copying file: file:/tmp/kv1.txt
Loading data to table default.pokes
rmr: DEPRECATED: Please use 'rm -r' instead.
Deleted /user/hive/warehouse/pokes
Table default.pokes stats: [num_partitions: 0, num_files: 1, num_rows: 0, total_size: 79, raw_data_size: 0]
OK
Time taken: 1.681 seconds

$ export HADOOP_MAPRED_HOME=/usr/lib/hadoop-mapreduce

4. 安装 Impala
$ cd /etc/yum.repos.d/
$ sudo wget http://archive.cloudera.com/impala/redhat/6/x86_64/impala/cloudera-impala.repo
$ sudo yum install impala impala-shell
$ sudo yum install impala-server impala-state-store

$ sudo vim /etc/hadoop/conf/hdfs-site.xml

...

   dfs.client.read.shortcircuit
   true


   dfs.domain.socket.path
   /var/run/hadoop-hdfs/dn._PORT
 

   dfs.client.file-block-storage-locations.timeout
   3000    


  dfs.datanode.hdfs-blocks-metadata.enabled
  true

$ sudo cp -rpa /etc/hadoop/conf/core-site.xml /etc/impala/conf/
$ sudo cp -rpa /etc/hadoop/conf/hdfs-site.xml /etc/impala/conf/

$ sudo service hadoop-hdfs-datanode restart

$ sudo service impala-state-store restart
$ sudo service impala-server restart

$ sudo /usr/java/default/bin/jps

5. 安装 Hbase
$ sudo yum install hbase

$ sudo vim /etc/security/limits.conf

hdfs - nofile 32768
hbase - nofile 32768

$ sudo vim /etc/pam.d/common-session

session required pam_limits.so

$ sudo vim /etc/hadoop/conf/hdfs-site.xml


  dfs.datanode.max.xcievers
  4096

$ sudo cp /usr/lib/impala/lib/hive-hbase-handler-0.10.0-cdh4.2.0.jar /usr/lib/hive/lib/hive-hbase-handler-0.10.0-cdh4.2.0.jar

$ sudo /etc/init.d/hadoop-hdfs-namenode restart
$ sudo /etc/init.d/hadoop-hdfs-datanode restart

$ sudo yum install hbase-master
$ sudo service hbase-master start

$ sudo -u hive hive

Logging initialized using configuration in file:/etc/hive/conf.dist/hive-log4j.properties
Hive history file=/tmp/hive/hive_job_log_hive_201305140905_2005531704.txt
hive> CREATE TABLE hbase_table_1(key int, value string) STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler' WITH SERDEPROPERTIES ("hbase.columns.mapping" = ":key,cf1:val") TBLPROPERTIES ("hbase.table.name" = "xyz");
OK
Time taken: 3.587 seconds

hive> INSERT OVERWRITE TABLE hbase_table_1 SELECT * FROM pokes WHERE foo=5;
Total MapReduce jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1368502088579_0004, Tracking URL = http://ip-10-197-10-4:8088/proxy/application_1368502088579_0004/
Kill Command = /usr/lib/hadoop/bin/hadoop job  -kill job_1368502088579_0004
Hadoop job information for Stage-0: number of mappers: 1; number of reducers: 0
2013-05-14 09:12:45,340 Stage-0 map = 0%,  reduce = 0%
2013-05-14 09:12:53,165 Stage-0 map = 100%,  reduce = 0%, Cumulative CPU 2.63 sec
MapReduce Total cumulative CPU time: 2 seconds 630 msec
Ended Job = job_1368502088579_0004
1 Rows loaded to hbase_table_1
MapReduce Jobs Launched: 
Job 0: Map: 1   Cumulative CPU: 2.63 sec   HDFS Read: 288 HDFS Write: 0 SUCCESS
Total MapReduce CPU Time Spent: 2 seconds 630 msec
OK
Time taken: 21.063 seconds

hive> select * from hbase_table_1;
OK
5	heylinx.com
Time taken: 0.685 seconds

hive> SELECT COUNT (*) FROM pokes;
Total MapReduce jobs = 1
Launching Job 1 out of 1
Number of reduce tasks determined at compile time: 1
In order to change the average load for a reducer (in bytes):
  set hive.exec.reducers.bytes.per.reducer=<number>
In order to limit the maximum number of reducers:
  set hive.exec.reducers.max=<number>
In order to set a constant number of reducers:
  set mapred.reduce.tasks=<number>
Starting Job = job_1368502088579_0005, Tracking URL = http://ip-10-197-10-4:8088/proxy/application_1368502088579_0005/
Kill Command = /usr/lib/hadoop/bin/hadoop job  -kill job_1368502088579_0005
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 1
2013-05-14 10:32:04,711 Stage-1 map = 0%,  reduce = 0%
2013-05-14 10:32:11,461 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.22 sec
2013-05-14 10:32:12,554 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.22 sec
2013-05-14 10:32:13,642 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.22 sec
2013-05-14 10:32:14,760 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.22 sec
2013-05-14 10:32:15,918 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.22 sec
2013-05-14 10:32:16,991 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.22 sec
2013-05-14 10:32:18,111 Stage-1 map = 100%,  reduce = 0%, Cumulative CPU 1.22 sec
2013-05-14 10:32:19,188 Stage-1 map = 100%,  reduce = 100%, Cumulative CPU 4.04 sec
MapReduce Total cumulative CPU time: 4 seconds 40 msec
Ended Job = job_1368502088579_0005
MapReduce Jobs Launched: 
Job 0: Map: 1  Reduce: 1   Cumulative CPU: 4.04 sec   HDFS Read: 288 HDFS Write: 2 SUCCESS
Total MapReduce CPU Time Spent: 4 seconds 40 msec
OK
5
Time taken: 28.195 seconds

6. 测试Impala性能
View parameters on http://ec2-204-236-182-78.us-west-1.compute.amazonaws.com:25000

$ impala-shell

[ip-10-197-10-4.us-west-1.compute.internal:21000] > CREATE TABLE IF NOT EXISTS pokes ( foo INT,bar STRING ) ROW FORMAT DELIMITED FIELDS TERMINATED BY "\t" LINES TERMINATED BY "\n";
Query: create TABLE IF NOT EXISTS pokes ( foo INT,bar STRING ) ROW FORMAT DELIMITED FIELDS TERMINATED BY "\t" LINES TERMINATED BY "\n"
[ip-10-197-10-4.us-west-1.compute.internal:21000] > show tables;
Query: show tables
Query finished, fetching results ...
+-------+
| name  |
+-------+
| pokes |
+-------+
Returned 1 row(s) in 0.00s

[ip-10-197-10-4.us-west-1.compute.internal:21000] > SELECT * from pokes;
Query: select * from pokes
Query finished, fetching results ...
+-----+-----------------+
| foo | bar             |
+-----+-----------------+
| 1   | www.baidu.com   |
| 2   | www.google.com  |
| 3   | www.sina.com.cn |
| 4   | www.163.com     |
| 5   | heylinx.com     |
+-----+-----------------+
Returned 5 row(s) in 0.28s

[ip-10-197-10-4.us-west-1.compute.internal:21000] > SELECT COUNT (*) from pokes;
Query: select COUNT (*) from pokes
Query finished, fetching results ...
+----------+
| count(*) |
+----------+
| 5        |
+----------+
Returned 1 row(s) in 0.34s

通过两个COUNT的结果来看,Hive使用了 28.195 seconds 而 Impala仅使用了0.34s,由此可以看出Impala的性能确实要优于Hive。

, , ,

  1. #1 by muxueqz on 2013/05/24 - 08:08

    文章不错,我有个疑问,
    Impala应该比Hive的内存/磁盘占用要高吧?

(will not be published)
*