oracle数据接入clickhouse

oracle数据接入clickhouse

oracle数据接入包括两个方面：一是CDC，基于日志做数据变化的捕捉，包括增删改；二是增量数据的准实时导入，依赖于自增id或时间字段，相对于CDC而言部署较为简单，适用场景也仅适用于增量数据。在此仅介绍增量数据的接入。

一、clickhouse单机安装

1. 升级OpenSSL

   • rpm -Uvh openssl-1.0.2k-12.el7.x86_64.rpm

2. 安装Unixodbc

   • rpm -ivh unixODBC-2.3.1-11.el7.x86_64.rpm

3. 安装clickhouse

   • http://repo.yandex.ru/clickhouse/rpm/stable/x86_64/

   • rpm -ivh clickhouse*

二、oracle–>flume–>kafka

尝试使用kafka connect 中的jdbc source 连接oracle，一直连接失败。但是连接mysql是可以的，暂时未找到问题所在。

1. flume安装

   • （略）使用hdp中 Flume 1.5.2

2. kafka安装

   • （略）使用hdp中kafka 0.10.1

3. flume-ng-sql-source

   • 下载地址：https://github.com/keedio/flume-ng-sql-source.git，http://repo.red-soft.biz/repos/clickhouse/stable/el7/
   • 编译：
     
     • 为clickhouse写入方便，将默认分隔符由‘，’改为‘\t’；
     • 以后可考虑增加json格式。
   • 把flume-ng-sql-source-1.4.3.jar放到flume的lib目录下

4. oracle

   • 建表

     • 
create table flume_ng_sql_source (
id varchar2(32) primary key,
msg varchar2(32),
createTime date not null
);


   • 插入数据

     • 
insert into flume_ng_sql_source(id,msg,createTime) values('1','Test increment Data',to_date('2017-08-01 07:06:20','yyyy-mm-dd hh24:mi:ss'));
insert into flume_ng_sql_source(id,msg,createTime) values('2','Test increment Data',to_date('2017-08-02 07:06:20','yyyy-mm-dd hh24:mi:ss'));
insert into flume_ng_sql_source(id,msg,createTime) values('3','Test increment Data',to_date('2017-08-03 07:06:20','yyyy-mm-dd hh24:mi:ss'));
insert into flume_ng_sql_source(id,msg,createTime) values('4','Test increment Data',to_date('2017-08-04 07:06:20','yyyy-mm-dd hh24:mi:ss'));
insert into flume_ng_sql_source(id,msg,createTime) values('5','Test increment Data',to_date('2017-08-05 07:06:20','yyyy-mm-dd hh24:mi:ss'));
insert into flume_ng_sql_source(id,msg,createTime) values('6','Test increment Data',to_date('2017-08-06 07:06:20','yyyy-mm-dd hh24:mi:ss'));
commit;


   • 把ojdbc6.jar放到flume的lib目录下

5. 新建flume-sql.conf

   • 在/usr/local/flume目录新建flume-sql.conf

     agentTest.channels = channelTest
     agentTest.sources = sourceTest
     agentTest.sinks = sinkTest
     
     ###########sql source#################
     
     
     # For each Test of the sources, the type is defined
     
     agentTest.sources.sourceTest.type = org.keedio.flume.source.SQLSource
     agentTest.sources.sourceTest.hibernate.connection.url = jdbc:oracle:thin:@10.8.7.96:1521/ora11g
     
     # Hibernate Database connection properties
     
     agentTest.sources.sourceTest.hibernate.connection.user = taizhou
     agentTest.sources.sourceTest.hibernate.connection.password = 123456
     agentTest.sources.sourceTest.hibernate.connection.autocommit = true
     agentTest.sources.sourceTest.hibernate.dialect = org.hibernate.dialect.Oracle10gDialect
     agentTest.sources.sourceTest.hibernate.connection.driver_class = oracle.jdbc.driver.OracleDriver
     agentTest.sources.sourceTest.run.query.delay=10000
     agentTest.sources.sourceTest.enclose.by.quotes = false
     
     agentTest.sources.sourceTest.status.file.path = /usr/local/flume
     agentTest.sources.sourceTest.status.file.name = agentTest.sqlSource.status
     
     # Custom query
     
     agentTest.sources.sourceTest.start.from = 2017-07-31 07:06:20
     agentTest.sources.sourceTest.custom.query = SELECT TO_CHAR(CREATETIME,'YYYY-MM-DD HH24:MI:SS'),ID,MSG FROM FLUME_NG_SQL_SOURCE WHERE CREATETIME > TO_DATE('$@$','YYYY-MM-DD HH24:MI:SS') ORDER BY CREATETIME ASC
     
     agentTest.sources.sourceTest.batch.size = 1000
     agentTest.sources.sourceTest.max.rows = 1000
     agentTest.sources.sourceTest.hibernate.connection.provider_class = org.hibernate.connection.C3P0ConnectionProvider
     agentTest.sources.sourceTest.hibernate.c3p0.min_size=1
     agentTest.sources.sourceTest.hibernate.c3p0.max_size=10
     
     ##############################
     
     agentTest.channels.channelTest.type = memory
     agentTest.channels.channelTest.capacity = 1000
     agentTest.channels.channelTest.transactionCapacity = 1000
     agentTest.channels.channelTest.byteCapacityBufferPercentage = 20
     agentTest.channels.channelTest.byteCapacity = 1600000
     
     agentTest.sinks.sinkTest.type = org.apache.flume.sink.kafka.KafkaSink
     agentTest.sinks.sinkTest.topic = test13
     agentTest.sinks.sinkTest.brokerList = 10.8.7.85:6667
     agentTest.sinks.sinkTest.requiredAcks = 1
     agentTest.sinks.sinkTest.batchSize = 20
     agentTest.sinks.sinkTest.channel = channelTest
     
     agentTest.sinks.sinkTest.channel = channelTest
     agentTest.sources.sourceTest.channels=channelTest

   • 注意时间字段为第一个；

   • $@$带引号，start.from值不需要引号，因其值填充$@$之后会有引号。

6. 启动flume

   • flume/bin目录下
     
     • flume-ng agent –conf conf –conf-file /usr/local/flume/flume-sql.conf –name agentTest -Dflume.root.logger=INFO,console
   • 查看是否有数据写入kafka
     
     • kafka-console-consumer.sh –zookeeper localhost:2181 –topic TestTopic
     • 数据为制表符分割，无引号

7. 查看状态文件

   • /usr/local/flume/agentTest.sqlSource.status
     
     • 其中LastIndex 值即为最后导入的最大时间字段
     • 若想从头重新导入，需把此文件删除

三、oracle–>flume–>clickhouse

flume的source及channel同接入kafka相同，仅需修改sink即可。

1. flume-clickhouse-sink

   • https://reviews.apache.org/r/50692/diff/1#2
   • 编译
   • flume-clickhouse-sink-1.5.2.jar放到flume的lib目录下

2. clickhouse建表

   • 开放远程访问

     • /etc/clickhouse-server/config.xml

   • <listen_host>::1</listen_host>
     <listen_host>本机ip</listen_host>

     clickhouse-client -m

     • CREATE TABLE flume_ng_sql_source
       (
       createtime DateTime,
       id UInt32, 
       msg String
       )  engine = MergeTree PARTITION BY toYYYYMMDD(createtime) order by id SETTINGS index_granularity = 8192;

     • 注意字段顺序同custom.query中相同

     • DateTime与Date区别

3. 新建ch.conf

   • 在/usr/local/flume目录新建ch.conf

   • agentTest.channels = channelTest
     agentTest.sources = sourceTest
     agentTest.sinks = sinkTest
     
     ###########sql source#################
     
     
     # For each Test of the sources, the type is defined
     
     agentTest.sources.sourceTest.type = org.keedio.flume.source.SQLSource
     agentTest.sources.sourceTest.hibernate.connection.url = jdbc:oracle:thin:@10.8.7.96:1521/ora11g
     
     # Hibernate Database connection properties
     
     agentTest.sources.sourceTest.hibernate.connection.user = taizhou
     agentTest.sources.sourceTest.hibernate.connection.password = 123456
     agentTest.sources.sourceTest.hibernate.connection.autocommit = true
     agentTest.sources.sourceTest.hibernate.dialect = org.hibernate.dialect.Oracle10gDialect
     agentTest.sources.sourceTest.hibernate.connection.driver_class = oracle.jdbc.driver.OracleDriver
     agentTest.sources.sourceTest.run.query.delay=10000
     agentTest.sources.sourceTest.enclose.by.quotes = false
     
     agentTest.sources.sourceTest.status.file.path = /usr/local/flume
     agentTest.sources.sourceTest.status.file.name = agentTest.sqlSource.status
     
     # Custom query
     
     agentTest.sources.sourceTest.start.from = 2017-07-31 07:06:20
     agentTest.sources.sourceTest.custom.query = SELECT TO_CHAR(CREATETIME,'YYYY-MM-DD HH24:MI:SS'),ID,MSG FROM FLUME_NG_SQL_SOURCE WHERE CREATETIME > TO_DATE('$@$','YYYY-MM-DD HH24:MI:SS') ORDER BY CREATETIME ASC
     
     agentTest.sources.sourceTest.batch.size = 1000
     agentTest.sources.sourceTest.max.rows = 1000
     agentTest.sources.sourceTest.hibernate.connection.provider_class = org.hibernate.connection.C3P0ConnectionProvider
     agentTest.sources.sourceTest.hibernate.c3p0.min_size=1
     agentTest.sources.sourceTest.hibernate.c3p0.max_size=10
     
     ##############################
     
     agentTest.channels.channelTest.type = memory
     agentTest.channels.channelTest.capacity = 1000
     agentTest.channels.channelTest.transactionCapacity = 1000
     agentTest.channels.channelTest.byteCapacityBufferPercentage = 20
     agentTest.channels.channelTest.byteCapacity = 1600000
     
     agentTest.sinks.sinkTest.type = org.apache.flume.sink.clickhouse.ClickHouseSink
     agentTest.sinks.sinkTest.host = http://10.8.7.96
     agentTest.sinks.sinkTest.port = 8123
     agentTest.sinks.sinkTest.database = default
     agentTest.sinks.sinkTest.table = flume_ng_sql_source
     agentTest.sinks.sinkTest.batchSize = 3000
     agentTest.sinks.sinkTest.format = TabSeparated
     
     agentTest.sinks.sinkTest.channel = channelTest
     agentTest.sources.sourceTest.channels=channelTest

4. 启动flume

   • flume/bin目录下
     
     • flume-ng agent –conf conf –conf-file /usr/local/flume/ch.conf –name agentTest -Dflume.root.logger=INFO,console

5. 查看clickhouse数据

   • 查询数据：
     
     • select * from flume_ng_sql_source order by id;
   • 查看数据目录：
     
     • /var/lib/clickhouse/data/default/flume_ng_sql_source/
     • 每个分区一个目录

6. 问题

   • 当数据从oracle读取成功，而写入clickhouse失败时，状态文件中的lastindex值也会改变。
   • 此流程是否应改为oracle–>flume–>kafka–>flume–>clickhouse

   参考：https://www.cnblogs.com/yangcx666/p/8723849.html

   ​