介绍impala的 rpm包安装形式，利用impala 2.2.0
impala 是一个数据查询引擎，类似hive，但跟hive解决的问题也有不太一样的地方，优点速度快，缺点吃内存，不适合大数据批处理
环境准备 注：本文的 hadoop hive hbase 都是通过tar 包独立安装的，并未通过rpm来安装，所以删除后面的软连接
1.下载rpm包 http://archive.cloudera.com/cdh5/redhat/6/x86_64/cdh/5.4.0/RPMS/x86_64/
共享我的百度云
1 2 3 4 5 6 7 8 bigtop-utils-0.7 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.47 .el6.noarch.rpm impala-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm impala-state-store-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm impala-shell-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm impala-catalog-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm impala-server-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm impala-udf-devel-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm impala-debuginfo-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm (可不装)
2.安装，建议上面的顺序
1 sudo rpm -ivh impala-2.2 .0 +cdh5.4.0 +0 -1 .cdh5.4.0 .p0.75 .el6.x86_64.rpm --nodeps --force
由于我的hadoop是通过tar包安装的，所以需要强制安装此rpm，其余正常安装
3.查看一下哪些有impala目录
1 2 3 4 5 6 7 8 9 [xiaolong@node007011 impala-rpm]$ sudo find / -name impala /var/log /impala /var/run/impala /var/lib/impala /var/lib/alternatives/impala /usr/lib/impala /etc/default/impala /etc/impala /etc/alternatives/impala
impala默认的安装目录为 /usr/lib/impala，jar包地址为/usr/lib/impala/lib/
4.删除默认的不存在软连接，并添加新连接。
1 2 3 4 5 6 7 8 9 10 11 sudo rm -rf /usr/lib/impala/lib/avro*.jar sudo rm -rf /usr/lib/impala/lib/hadoop-*.jar sudo rm -rf /usr/lib/impala/lib/hive-*.jar sudo rm -rf /usr/lib/impala/lib/hbase-*.jar sudo rm -rf /usr/lib/impala/lib/parquet-hadoop-bundle.jar sudo rm -rf /usr/lib/impala/lib/sentry-*.jar sudo rm -rf /usr/lib/impala/lib/zookeeper.jar sudo rm -rf /usr/lib/impala/lib/libhadoop.so sudo rm -rf /usr/lib/impala/lib/libhadoop.so.1.0 .0 sudo rm -rf /usr/lib/impala/lib/libhdfs.so sudo rm -rf /usr/lib/impala/lib/libhdfs.so.0.0 .0
5.建立新连接，建议利用脚本运行，自己设一下各个home即可，软连后的名字不能改。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 sudo ln -s $HBASE_HOME /lib/avro-1.7 .6 -cdh5.4.0 .jar /usr/lib/impala/lib/avro.jar sudo ln -s $HADOOP_HOME /share/hadoop/common/lib/hadoop-annotations-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-annotations.jar sudo ln -s $HADOOP_HOME /share/hadoop/common/lib/hadoop-auth-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-auth.jar sudo ln -s $HADOOP_HOME /share/hadoop/tools/lib/hadoop-aws-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-aws.jar sudo ln -s $HADOOP_HOME /share/hadoop/common/hadoop-common-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-common.jar sudo ln -s $HADOOP_HOME /share/hadoop/hdfs/hadoop-hdfs-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-hdfs.jar sudo ln -s $HADOOP_HOME /share/hadoop/mapreduce/hadoop-mapreduce-client-common-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-mapreduce-client-common.jar sudo ln -s $HADOOP_HOME /share/hadoop/mapreduce/hadoop-mapreduce-client-core-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-mapreduce-client-core.jar sudo ln -s $HADOOP_HOME /share/hadoop/mapreduce/hadoop-mapreduce-client-jobclient-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-mapreduce-client-jobclient.jar sudo ln -s $HADOOP_HOME /share/hadoop/mapreduce/hadoop-mapreduce-client-shuffle-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-mapreduce-client-shuffle.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-api-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-api.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-client-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-client.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-common-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-common.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-server-applicationhistoryservice-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-server-applicationhistoryservice.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-server-common-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-server-common.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-server-nodemanager-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-server-nodemanager.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-server-resourcemanager-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-server-resourcemanager.jar sudo ln -s $HADOOP_HOME /share/hadoop/yarn/hadoop-yarn-server-web-proxy-2.6 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hadoop-yarn-server-web-proxy.jar sudo ln -s $HBASE_HOME /lib/hbase-annotations-1.0 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hbase-annotations.jar sudo ln -s $HBASE_HOME /lib/hbase-client-1.0 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hbase-client.jar sudo ln -s $HBASE_HOME /lib/hbase-common-1.0 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hbase-common.jar sudo ln -s $HBASE_HOME /lib/hbase-protocol-1.0 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hbase-protocol.jar sudo ln -s $HIVE_HOME /lib/hive-ant-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-ant.jar sudo ln -s $HIVE_HOME /lib/hive-beeline-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-beeline.jar sudo ln -s $HIVE_HOME /lib/hive-common-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-common.jar sudo ln -s $HIVE_HOME /lib/hive-exec-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-exec.jar sudo ln -s $HIVE_HOME /lib/hive-hbase-handler-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-hbase-handler.jar sudo ln -s $HIVE_HOME /lib/hive-metastore-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-metastore.jar sudo ln -s $HIVE_HOME /lib/hive-serde-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-serde.jar sudo ln -s $HIVE_HOME /lib/hive-service-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-service.jar sudo ln -s $HIVE_HOME /lib/hive-shims-common-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-shims-common.jar sudo ln -s $HIVE_HOME /lib/hive-shims-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-shims.jar sudo ln -s $HIVE_HOME /lib/hive-shims-scheduler-1.1 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/hive-shims-scheduler.jar sudo ln -s $HADOOP_HOME /lib/native/libhadoop.so /usr/lib/impala/lib/libhadoop.so sudo ln -s $HADOOP_HOME /lib/native/libhadoop.so.1.0 .0 /usr/lib/impala/lib/libhadoop.so.1.0 .0 sudo ln -s $HADOOP_HOME /lib/native/libhdfs.so /usr/lib/impala/lib/libhdfs.so sudo ln -s $HADOOP_HOME /lib/native/libhdfs.so.0.0 .0 /usr/lib/impala/lib/libhdfs.so.0.0 .0 sudo ln -s $HIVE_HOME /lib/parquet-hadoop-bundle-1.5 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/parquet-hadoop-bundle.jar sudo ln -s $SENTRY_HOME /lib/sentry-binding-hive-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-binding-hive.jar sudo ln -s $SENTRY_HOME /lib/sentry-core-common-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-core-common.jar sudo ln -s $SENTRY_HOME /lib/sentry-core-model-db-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-core-model-db.jar sudo ln -s $SENTRY_HOME /lib/sentry-policy-common-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-policy-common.jar sudo ln -s $SENTRY_HOME /lib/sentry-policy-db-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-policy-db.jar sudo ln -s $SENTRY_HOME /lib/sentry-provider-cache-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-provider-cache.jar sudo ln -s $SENTRY_HOME /lib/sentry-provider-common-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-provider-common.jar sudo ln -s $SENTRY_HOME /lib/sentry-provider-db-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-provider-db.jar sudo ln -s $SENTRY_HOME /lib/sentry-provider-file-1.4 .0 -cdh5.4.0 .jar /usr/lib/impala/lib/sentry-provider-file.jar sudo ln -s $ZOOKEEPER_HOME /zookeeper-3.4 .5 -cdh5.3.3 .jar /usr/lib/impala/lib/zookeeper.jar
配置 1.修改/etc/default/bigtop-utils
1 export JAVA_HOME=/usr/local /datacenter/jdk1.7.0 _79
2.修改 /etc/default/impala 默认文件，除了修改环境，还需要修改启动参数，显示传递给 statestored 和 catalogd 它们各自需要的ip和port ，不然都默认 localhost ，分开部署会有问题的。
1 2 3 4 5 6 7 IMPALA_CATALOG_SERVICE_HOST=192.168 .7.11 IMPALA_STATE_STORE_HOST=192.168 .7.11 IMPALA_CATALOG_ARGS=" -log_dir=${IMPALA_LOG_DIR} -catalog_service_host=${IMPALA_CATALOG_SERVICE_HOST} -state_store_host=${IMPALA_STATE_STORE_HOST} " IMPALA_STATE_STORE_ARGS=" -log_dir=${IMPALA_LOG_DIR} -state_store_host=${IMPALA_STATE_STORE_HOST} -state_store_port=${IMPALA_STATE_STORE_PORT} -catalog_service_host=${IMPALA_CATALOG_SERVICE_HOST} -catalog_service_port=${CATALOG_SERVICE_PORT} " MYSQL_CONNECTOR_JAR=/usr/local /datacenter/hive/lib/mysql-connector-java-5.1 .34 .jar
3.根据实际环境修改impala相关脚本文件 /etc/init.d/impala-state-store、/etc/init.d/impala-server、/etc/init.d/impala-catalog，修改其中两处跟用户相关的地方
1 2 3 4 SVC_USER="hadoop" install -d -m 0755 -o hadoop -g hadoop /var/run/impala 1 >/dev/null 2 >&1 || :
为保险起见 手工建立日志目录并修改权限 mkdir /var/log/impala && chmod -R 777 /var/log/impala
4.修改 hdfs-site.xml 这一处就够了
hdfs-site.xml 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 <property > <name > dfs.client.read.shortcircuit</name > <value > true</value > </property > <property > <name > dfs.domain.socket.path</name > <value > /var/run/hadoop-hdfs/dn</value > </property > <property > <name > dfs.datanode.hdfs-blocks-metadata.enabled</name > <value > true</value > </property > <property > <name > dfs.client.use.legacy.blockreader.local</name > <value > false</value > ###编者注：这里需要为false ，官网文档错了 </property > <property > <name > dfs.datanode.data.dir.perm</name > <value > 750</value > </property > <property > <name > dfs.block.local-path-access.user</name > <value > hadoop</value > </property > <property > <name > dfs.client.file-block-storage-locations.timeout</name > <value > 3000</value > </property >
官网的文档有一处错误，dfs.client.use.legacy.blockreader.local 属性应该为 false，文档中是 true
http://www.cloudera.com/content/www/en-us/documentation/archive/impala/2-x/2-1-x/topics/impala_config_performance.html?scroll=config_performance
注意每个节点创建 hdfs-site.xml 里指定的 /var/run/hdfs-sockets/dn 权限
5.impalad的配置文件路径由环境变量IMPALA_CONF_DIR指定，默认为 /etc/impala/conf，拷贝配置或建立软连接 的hive-site.xml、core-site.xml、hdfs-site.xml、hbase-site.xml文件至/etc/impala/conf目录下。
1 2 3 4 sudo ln -s /usr/local /datacenter/hadoop/etc/hadoop/core-site.xml /etc/impala/conf/ sudo ln -s /usr/local /datacenter/hadoop/etc/hadoop/hdfs-site.xml /etc/impala/conf/ sudo ln -s /usr/local /datacenter/hive/conf/hive-site.xml /etc/impala/conf/ sudo ln -s /usr/local /datacenter/hbase/conf/hbase-site.xml /etc/impala/conf/
启动 启动是最诡异的地方，由于官网建议statestored和catalogd在同一台机器上，所以ip一样，但端口不一样，然后利用service启动就只能启动一个，可以通过 pid file 看到是同一个进程.
1.在 statestored 和 catalogd 机器上启动「impala主节点」， 建议启动时，一个利用service 启动一个利用手工启动
1 2 3 4 cd /usr/lib/impalanohup /usr/lib/impala/sbin/statestored -log_dir=/var/log /impala -state_store_host=192.168 .7.11 -state_store_port=24000 -catalog_service_host=192.168 .7.11 :26000 -catalog_service_port=26000 & sudo service impala-catalog start
碰见问题 ,缺少jdk so文件，缺少classpath impala ，所以修改环境变量
1 2 3 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH :/usr/local /lib:/usr/local /datacenter/jdk1.7.0 _79/jre/lib/amd64:/usr/local /datacenter/jdk1.7.0 _79/jre/lib/amd64/server:/usr/lib/impala/lib/export CLASSPATH=$CLASSPATH :$JAVA_HOME /lib:$JAVA_HOME /jre/lib:/usr/lib/impala/lib
2.在「impala从节点」上，启动 serverd
1 sudo service impala-server start
3.访问主节点 25010 端口，得到statestore web ui ，其中比较有参考意义的，logs 和 varz 里面有环境变量的信息
4.访问从节点的 25000 ，得到 serverd 的 web ui
执行 1.进入impala-shell，即连接的本机的 impala-server
1 2 3 4 [root@node007012 ~] Starting Impala Shell without Kerberos authentication Connected to node007012:21000 Server version: impalad version 2.2 .0 -cdh5 RELEASE (build 2 ffd73a4255cefd521362ffe1cfb37463f67f75c)
2.同步hive元信息，以便感知已存在的hive库表
1 2 3 4 [node007012:21000 ] > invalidate metadata; Query: invalidate metadata Fetched 0 row(s) in 3.98 s
3.执行一下ddl语句，看是否正常
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 [node007012:21000 ] > SHOW DATABASES; Query: show DATABASES +------------------+ | name | +------------------+ | _impala_builtins | | default | | test | +------------------+ Fetched 3 row(s) in 0.01 s [node007012:21000 ] > USE test ; Query: use test [node007012:21000 ] > show tables; Query: show tables +--------------+ | name | +--------------+ | g_info | | g_info_pa | | g_info_sqoop | | pokes | | pokes_pa | | sales_order | | t1 | | t2 | | user_sqoop | +--------------+ Fetched 9 row(s) in 0.01 s
4.验证查询语句，其中g_info表是textfile，g_info_pa 是parquetfile 的，可见效率差异性
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 [node007012:21000 ] > select count(*) from g_info; Query: select count(*) from g_info +----------+ | count(*) | +----------+ | 1222023 | +----------+ WARNINGS: Unknown disk id. This will negatively affect performance. Check your hdfs settings to enable block location metadata. (1 of 8 similar) Fetched 1 row(s) in 44.13 s [node007012:21000 ] > select count(*) from g_info_pa; Query: select count(*) from g_info_pa +----------+ | count(*) | +----------+ | 1192760 | +----------+ WARNINGS: Unknown disk id. This will negatively affect performance. Check your hdfs settings to enable block location metadata. (1 of 4 similar) Fetched 1 row(s) in 4.73 s
这个警告日志里报错的是，待查
1 2 3 W1226 19 :14 :35.052875 30015 DomainSocketFactory.java:167 ] error creating DomainSocket Java exception follows: java.net.ConnectException: connect(2 ) error: Connection refused when trying to connect to '/var/run/hdfs-sockets/dn'
参考：
