CDH（Cloudera's Distribution, including Apache Hadoop）是众多 Hadoop 发行版本中的一种。CDH的最新版本是6.0.1，支持Hadoop 3.0.0，本文介绍如何使 CDH6 的相关组件（Hadoop、Hive、Spark、Impala 等）能够查询 OSS。
前提条件
步骤一：增加 OSS 配置
- 通过集群管理工具 CM 来增加配置。
若没有 CM 管理的集群，可以修改 core-site.xml。以 CM 为例，需要增加如下配置：
更多参数解释请参见 Hadoop-Aliyun module。配置项 值 fs.oss.endpoint 填写需要连接的 OSS 的 Endpoint。 例如：oss-cn-zhangjiakou-internal.aliyuncs.com
fs.oss.accessKeyId 填写 OSS 的 AccessKeyId。 fs.oss.accessKeySecret 填写 OSS 的 AccessKeySecret。 fs.oss.impl Hadoop OSS 文件系统实现类。目前固定为：org.apache.hadoop.fs.aliyun.oss.AliyunOSSFileSystem fs.oss.buffer.dir 填写临时文件目录。 建议值：/tmp/oss
fs.oss.connection.secure.enabled 是否开启 HTTPS。开启HTTPS会影响性能。 建议值：false
fs.oss.connection.maximum 与 OSS 的连接数 建议值：2048
- 根据 CM 提示重启集群。
- 测试读写 OSS。
- 测试读：
hadoop fs -ls oss://${your-bucket-name}/ - 测试写：
hadoop fs -mkdir oss://${your-bucket-name}/hadoop-test若测试可以读写 OSS，则配置成功；若无法读写 OSS，请检查配置。
说明 本文中所有 ${} 的内容为环境变量，请根据您实际的环境修改。
- 测试读：
步骤二：配置 Apache Impala 对 OSS 的支持
虽然 CDH6 支持 OSS，但是它里面的 Impala 组件却没有将 OSS 的支持放到它的 CLASSPATH 里面去，因此您需要在所有的 Impala 节点执行如下操作：
- 进入 ${CDH_HOME}/lib/impala 目录，创建符号链接：
[root@cdh-master impala]# cd lib/ [root@cdh-master lib]# ln -s ../../../jars/hadoop-aliyun-3.0.0-cdh6.0.1.jar hadoop-aliyun.jar [root@cdh-master lib]# ln -s ../../../jars/aliyun-sdk-oss-2.8.3.jar aliyun-sdk-oss-2.8.3.jar [root@cdh-master lib]# ln -s ../../../jars/jdom-1.1.jar jdom-1.1.jar
- 进入${CDH_HOME}/bin目录，修改impalad、statestored、catalogd这三个文件，在文件最后一行exec命令前增加如下内容：
export CLASSPATH=${CLASSPATH}:${IMPALA_HOME}/lib/hadoop-aliyun.jar:${IMPALA_HOME}/lib/aliyun-sdk-oss-2.8.3.jar:${IMPALA_HOME}/lib/jdom-1.1.jar - 重启所有节点的 impala 相关进程。
验证配置
尝试使用 Apache Impala 查询 OSS，并且运行 TPC-DS 的查询语句。相关介绍请参见 Apache Impala 的介绍和TPC-DS查询。
TPC-DS 基于 Hive QL，因为 Hive QL与 Impala SQL兼容性比较高，所以我们用这个作为实验案例（有几个 query 因为兼容性问题不能运行）：
- 生成 sample 数据到 OSS。
[root@cdh-master ~]# git clone https://github.com/hortonworks/hive-testbench.git [root@cdh-master ~]# cd hive-testbench [root@cdh-master hive-testbench]# git checkout hive14 [root@cdh-master hive-testbench]# ./tpcds-build.sh [root@cdh-master hive-testbench]# FORMAT=textfile ./tpcds-setup.sh 50 oss://{your-bucket-name}/ - 建表。
这个 benchmark 的建表语句与 Apache Impala 的建表语句兼容，复制 ddl-tpcds/text/alltables.sql中的建表语句，修改 ${LOCATION}即可，例如：
[root@cdh-master hive-testbench]# impala-shell -i cdh-slave01 -d default Starting Impala Shell without Kerberos authentication Connected to cdh-slave01:21000 Server version: impalad version 3.0.0-cdh6.0.1 RELEASE(build9a74a5053de5f7b8dd983802e6d75e58d31472db) *********************************************************************************** Welcome to the Impala shell. (Impala Shell v3.0.0-cdh6.0.1(9a74a50) built on Wed Sep 1911:27:37 PDT 2018) Want to know what version of Impala you're connected to? Run the VERSION command to find out! *********************************************************************************** Query: use `default` [cdh-slave01:21000] default> create external table call_center( > cc_call_center_sk bigint > , cc_call_center_id string > , cc_rec_start_date string > , cc_rec_end_date string > , cc_closed_date_sk bigint > , cc_open_date_sk bigint > , cc_name string > , cc_class string > , cc_employees int > , cc_sq_ft int > , cc_hours string > , cc_manager string > , cc_mkt_id int > , cc_mkt_class string > , cc_mkt_desc string > , cc_market_manager string > , cc_division int > , cc_division_name string > , cc_company int > , cc_company_name string > , cc_street_number string > , cc_street_name string > , cc_street_type string > , cc_suite_number string > , cc_city string > , cc_county string > , cc_state string > , cc_zip string > , cc_country string > , cc_gmt_offset double > , cc_tax_percentage double > ) > row format delimited fields terminated by '|' > location 'oss://{your-bucket-name}/50/call_center'; Query: create external table call_center( cc_call_center_sk bigint , cc_call_center_id string , cc_rec_start_date string , cc_rec_end_date string , cc_closed_date_sk bigint , cc_open_date_sk bigint , cc_name string , cc_class string , cc_employees int , cc_sq_ft int , cc_hours string , cc_manager string , cc_mkt_id int , cc_mkt_class string , cc_mkt_desc string , cc_market_manager string , cc_division int , cc_division_name string , cc_company int , cc_company_name string , cc_street_number string , cc_street_name string , cc_street_type string , cc_suite_number string , cc_city string , cc_county string , cc_state string , cc_zip string , cc_country string , cc_gmt_offset double , cc_tax_percentage double ) row format delimited fields terminated by '|' location 'oss://{your-bucket-name}/50/call_center' +-------------------------+ | summary | +-------------------------+ | Table has been created. | +-------------------------+ Fetched 1 row(s) in 4.10s - 检查建表内容。
[cdh-slave01:21000] default> show tables; Query: show tables +------------------------+ | name | +------------------------+ | call_center | | catalog_page | | catalog_returns | | catalog_sales | | customer | | customer_address | | customer_demographics | | date_dim | | household_demographics | | income_band | | inventory | | item | | promotion | | reason | | ship_mode | | store | | store_returns | | store_sales | | time_dim | | warehouse | | web_page | | web_returns | | web_sales | | web_site | +------------------------+ Fetched 24 row(s) in 0.03s
- 在 Impala 上执行 TPC-DS 的查询，已经可以正常查询 OSS 的数据。
- 查询的 SQL 文件在 sample-queries-tpcds 目录下，进入该目录。
[root@cdh-master hive-testbench]# ls sample-queries-tpcds query12.sql query20.sql query27.sql query39.sql query46.sql query54.sql query64.sql query71.sql query7.sql query87.sql query93.sql README.md query13.sql query21.sql query28.sql query3.sql query48.sql query55.sql query65.sql query72.sql query80.sql query88.sql query94.sql testbench.settings query15.sql query22.sql query29.sql query40.sql query49.sql query56.sql query66.sql query73.sql query82.sql query89.sql query95.sql testbench-withATS.settings query17.sql query24.sql query31.sql query42.sql query50.sql query58.sql query67.sql query75.sql query83.sql query90.sql query96.sql query18.sql query25.sql query32.sql query43.sql query51.sql query60.sql query68.sql query76.sql query84.sql query91.sql query97.sql query19.sql query26.sql query34.sql query45.sql query52.sql query63.sql query70.sql query79.sql query85.sql query92.sql query98.sql
- 执行 TPC-DS 查询 query13.sql。
[root@cdh-master hive-testbench]# impala-shell -i cdh-slave01 -d default -f sample-queries-tpcds/query13.sql Starting Impala Shell without Kerberos authentication Connected to cdh-slave01:21000 Server version: impalad version 3.0.0-cdh6.0.1 RELEASE(build9a74a5053de5f7b8dd983802e6d75e58d31472db) Query: use`default`Query: selectavg(ss_quantity) ,avg(ss_ext_sales_price) ,avg(ss_ext_wholesale_cost) ,sum(ss_ext_wholesale_cost) from store_sales ,store ,customer_demographics ,household_demographics ,customer_address ,date_dim where store.s_store_sk = store_sales.ss_store_sk and store_sales.ss_sold_date_sk = date_dim.d_date_sk and date_dim.d_year = 2001and((store_sales.ss_hdemo_sk=household_demographics.hd_demo_sk and customer_demographics.cd_demo_sk = store_sales.ss_cdemo_sk and customer_demographics.cd_marital_status = 'M'and customer_demographics.cd_education_status = '4 yr Degree'and store_sales.ss_sales_price between100.00and150.00and household_demographics.hd_dep_count = 3 )or (store_sales.ss_hdemo_sk=household_demographics.hd_demo_sk and customer_demographics.cd_demo_sk = store_sales.ss_cdemo_sk and customer_demographics.cd_marital_status = 'D'and customer_demographics.cd_education_status = 'Primary'and store_sales.ss_sales_price between50.00and100.00and household_demographics.hd_dep_count = 1 ) or (store_sales.ss_hdemo_sk=household_demographics.hd_demo_sk and customer_demographics.cd_demo_sk = ss_cdemo_sk and customer_demographics.cd_marital_status = 'U'and customer_demographics.cd_education_status = 'Advanced Degree'and store_sales.ss_sales_price between150.00and200.00and household_demographics.hd_dep_count = 1 )) and((store_sales.ss_addr_sk = customer_address.ca_address_sk and customer_address.ca_country = 'United States'and customer_address.ca_state in('KY', 'GA', 'NM') and store_sales.ss_net_profit between100and200 ) or (store_sales.ss_addr_sk = customer_address.ca_address_sk and customer_address.ca_country = 'United States'and customer_address.ca_state in('MT', 'OR', 'IN') and store_sales.ss_net_profit between150and300 ) or (store_sales.ss_addr_sk = customer_address.ca_address_sk and customer_address.ca_country = 'United States'and customer_address.ca_state in('WI', 'MO', 'WV') and store_sales.ss_net_profit between50and250 )) Query submitted at: 2018-10-3011:44:47(Coordinator: http://cdh-slave01:25000) Query progress can be monitored at: http://cdh-slave01:25000/query_plan?query_id=ff4b3157eddfc3c4:8a31c6500000000 +-------------------+-------------------------+----------------------------+----------------------------+ | avg(ss_quantity) | avg(ss_ext_sales_price) | avg(ss_ext_wholesale_cost) | sum(ss_ext_wholesale_cost) | +-------------------+-------------------------+----------------------------+----------------------------+ | 30.87106918238994 | 2352.642327044025 | 2162.600911949685 | 687707.09 | +-------------------+-------------------------+----------------------------+----------------------------+ Fetched 1 row(s) in 353.16s
本次查询涉及到 6 张表： store、 store_sales、 customer_demographics、 household_demographics、 customer_address、 date_dim。[cdh-slave01:21000] default> showtable STATS store; Query: showtable STATS store +-------+--------+---------+--------------+-------------------+--------+-------------------+-------------------------------------------+ | #Rows | #Files | Size | Bytes Cached | CacheReplication | Format | Incremental stats | Location | +-------+--------+---------+--------------+-------------------+--------+-------------------+-------------------------------------------+ | -1 | 1 | 37.56KB | NOT CACHED | NOT CACHED | TEXT | false | oss://{your-bucket-name}/50/store | +-------+--------+---------+--------------+-------------------+--------+-------------------+-------------------------------------------+ Fetched 1 row(s) in 0.01s [cdh-slave01:21000] default> showtable STATS store_sales; Query: showtable STATS store_sales +-------+--------+---------+--------------+-------------------+--------+-------------------+-------------------------------------------------+ | #Rows | #Files | Size | Bytes Cached | CacheReplication | Format | Incremental stats | Location | +-------+--------+---------+--------------+-------------------+--------+-------------------+-------------------------------------------------+ | -1 | 50 | 18.75GB | NOT CACHED | NOT CACHED | TEXT | false | oss://{your-bucket-name}/50/store_sales | +-------+--------+---------+--------------+-------------------+--------+-------------------+-------------------------------------------------+ Fetched 1 row(s) in 0.01s [cdh-slave01:21000] default> showtable STATS customer_demographics; Query: showtable STATS customer_demographics +-------+--------+---------+--------------+-------------------+--------+-------------------+-----------------------------------------------------------+ | #Rows | #Files | Size | Bytes Cached | CacheReplication | Format | Incremental stats | Location | +-------+--------+---------+--------------+-------------------+--------+-------------------+-----------------------------------------------------------+ | -1 | 50 | 76.92MB | NOT CACHED | NOT CACHED | TEXT | false | oss://{your-bucket-name}/50/customer_demographics | +-------+--------+---------+--------------+-------------------+--------+-------------------+-----------------------------------------------------------+ Fetched 1 row(s) in 0.01s [cdh-slave01:21000] default> showtable STATS household_demographics; Query: showtable STATS household_demographics +-------+--------+----------+--------------+-------------------+--------+-------------------+------------------------------------------------------------+ | #Rows | #Files | Size | Bytes Cached | CacheReplication | Format | Incremental stats | Location | +-------+--------+----------+--------------+-------------------+--------+-------------------+------------------------------------------------------------+ | -1 | 1 | 148.10KB | NOT CACHED | NOT CACHED | TEXT | false | oss://{your-bucket-name}/50/household_demographics | +-------+--------+----------+--------------+-------------------+--------+-------------------+------------------------------------------------------------+ Fetched 1 row(s) in 0.01s [cdh-slave01:21000] default> showtable STATS customer_address; Query: showtable STATS customer_address +-------+--------+---------+--------------+-------------------+--------+-------------------+------------------------------------------------------+ | #Rows | #Files | Size | Bytes Cached | CacheReplication | Format | Incremental stats | Location | +-------+--------+---------+--------------+-------------------+--------+-------------------+------------------------------------------------------+ | -1 | 1 | 40.54MB | NOT CACHED | NOT CACHED | TEXT | false | oss://{your-bucket-name}/50/customer_address | +-------+--------+---------+--------------+-------------------+--------+-------------------+------------------------------------------------------+ Fetched 1 row(s) in 0.01s [cdh-slave01:21000] default> showtable STATS date_dim; Query: showtable STATS date_dim +-------+--------+--------+--------------+-------------------+--------+-------------------+----------------------------------------------+ | #Rows | #Files | Size | Bytes Cached | CacheReplication | Format | Incremental stats | Location | +-------+--------+--------+--------------+-------------------+--------+-------------------+----------------------------------------------+ | -1 | 1 | 9.84MB | NOT CACHED | NOT CACHED | TEXT | false | oss://{your-bucket-name}/50/date_dim | +-------+--------+--------+--------------+-------------------+--------+-------------------+----------------------------------------------+ Fetched 1 row(s) in 0.01s - 查询的 SQL 文件在 sample-queries-tpcds 目录下，进入该目录。
参考文档
更多内容，请参见如下文档：
您也可以通过阿里云 EMR 访问 OSS。阿里云 EMR 基于开源生态，包括 Hadoop、Spark、Kafka、Flink、Storm 等组件，为您提供集群、作业、数据管理等服务的一站式企业大数据平台，并无缝支持 OSS。阿里云 EMR 与 OSS 紧密结合，针对开源生态访问 OSS，有多项技术优化，详情请参见 EMR 产品介绍。