flume mysql 导入数据到 hdfs

#声明source, channel, sink
a1.sources=sqlSource
a1.channels=c1
a1.sinks=s1

#声明source类型
a1.sources.sqlSource.type=org.keedio.flume.source.SQLSource
a1.sources.sqlSource.hibernate.connection.url=jdbc:mysql://10.1.20.115:3306/t_datax
a1.sources.sqlSource.hibernate.connection.user=root
a1.sources.sqlSource.hibernate.connection.password=123456

#这个参数很重要，任务自动提交，默认为false,如果不设置为true,查询不会自动执行
a1.sources.sqlSource.hibernate.connection.autocommit=true
#声明mysql的hibernate方言
a1.sources.sqlSource.hibernate.dialect= org.hibernate.dialect.MySQL5Dialect
a1.sources.sqlSource.hibernate.connection.driver_calss=com.mysql.jdbc.Driver

#查询时间间隔
a1.sources.sqlSource.run.query..delay=10000

#声明保存flume状态的文件夹位置
a1.sources.sqlSource.status.file.path=/opt/flume/log
a1.sources.sqlSource.status.file.name=sqlSource.status

#声明查询开始位置
a1.sources.sqlSource.start.from=0

#sql语句自定义,但是要注意:增量只能针对id字段即主键列,经测试系统默认如此.
#而且必须要将主键查询出来,因为如果不查询主键,flume无法记录上一次查询的位置.
#$@$表示增量列上一次查询的值，记录在status文件中
a1.sources.sqlSource.custom.query=select id, name, img, utime from t_datax1 where id > $@$

#设置分配参数
a1.sources.sqlSource.batch.size=1000
a1.sources.sqlSource.max.rows=1000

#查询结果分隔符
a1.sources.sqlSource.delimiter.entty=|

#a1.sources.sqlSource.hibernate.connection.provider_class = org.hibernate.connection.C3P0ConnectionProvider
#a1.sources.sqlSource.hibernate.c3p0.min_size=3
#a1.sources.sqlSource.hibernate.c3p0.max_size=10


a1.sources.sqlSource.interceptors=i1
a1.sources.sqlSource.interceptors.i1.type=search_replace
a1.sources.sqlSource.interceptors.i1.searchPattern="
a1.sources.sqlSource.interceptors.i1.replaceString=

#设置通道为内存模式
a1.channels.c1.type=memory
a1.channels.c1.capacity=10000
a1.channels.c1.transactionCapacity=10000
a1.channels.c1.byteCapacityBufferPercentage=20
a1.channels.c1.byteCapacity=800000


a1.sinks.s1.type=HDFS
#ns为namenode的命名空间,两个作用,一个是防止集群坍塌,另一个是改参数只能作用在active的namenode节点上
a1.sinks.s1.hdfs.path=hdfs://10.1.80.5:8020/user/root/test
a1.sinks.s1.hdfs.fileType=DataStream
a1.sinks.s1.hdfs.writeFormat=Text
#设置滚动时间,每隔多少时间生成一个文件.如果设置成0,则禁止滚动,可以使所有数据被写到一个文件中.
a1.sinks.s1.hdfs.rollInterval=0
#设置文件存储数据多大的时候生成下一个文件,建议设置成128M和块大小相同
a1.sinks.s1.hdfs.rollSize=134217728
#设置文件多少行时,滚动生成下一个文件,设置成0时禁止滚动
a1.sinks.s1.hdfs.rollCount=0

a1.sources.sqlSource.channels=c1
a1.sinks.s1.channel=c1

flume 增量导入数据到hdfs:

环境需求：

​ jdk1.8

​ maven

插件：flume-ng-sql-source

​ https://github.com/keedio/flume-ng-sql-source

flume-ng-sql-source

This project is used for flume-ng to communicate with sql databases

Current sql database engines supported

• After the last update the code has been integrated with hibernate, so all databases supported by this technology should work.

Compilation and packaging

  $ mvn package

Deployment

Copy flume-ng-sql-source-.jar in target folder into flume plugins dir folder

  $ mkdir -p $FLUME_HOME/plugins.d/sql-source/lib $FLUME_HOME/plugins.d/sql-source/libext
  $ cp flume-ng-sql-source-0.8.jar $FLUME_HOME/plugins.d/sql-source/lib

Specific installation by database engine

MySQL

Download the official mysql jdbc driver and copy in libext flume plugins directory:

$ wget https://dev.mysql.com/get/Downloads/Connector-J/mysql-connector-java-5.1.35.tar.gz
$ tar xzf mysql-connector-java-5.1.35.tar.gz
$ cp mysql-connector-java-5.1.35-bin.jar $FLUME_HOME/plugins.d/sql-source/libext

Microsoft SQLServer

Download the official Microsoft 4.1 Sql Server jdbc driver and copy in libext flume plugins directory:
Download URL: https://www.microsoft.com/es-es/download/details.aspx?id=11774

$ tar xzf sqljdbc_4.1.5605.100_enu.tar.gz
$ cp sqljdbc_4.1/enu/sqljdbc41.jar $FLUME_HOME/plugins.d/sql-source/libext

IBM DB2

Download the official IBM DB2 jdbc driver and copy in libext flume plugins directory: Download URL: http://www-01.ibm.com/support/docview.wss?uid=swg21363866

Configuration of SQL Source:

Mandatory properties in bold

Property Name	Default	Description
channels	-	Connected channel names
type	-	The component type name, needs to be org.keedio.flume.source.SQLSource
hibernate.connection.url	-	Url to connect with the remote Database
hibernate.connection.user	-	Username to connect with the database
hibernate.connection.password	-	Password to connect with the database
table	-	Table to export data
status.file.name	-	Local file name to save last row number read
status.file.path	/var/lib/flume	Path to save the status file
start.from	0	Start value to import data
delimiter.entry	,	delimiter of incoming entry
enclose.by.quotes	true	If Quotes are applied to all values in the output.
columns.to.select	*	Which colums of the table will be selected
run.query.delay	10000	ms to wait between run queries
batch.size	100	Batch size to send events to flume channel
max.rows	10000	Max rows to import per query
read.only	false	Sets read only session with DDBB
custom.query	-	Custom query to force a special request to the DB, be carefull. Check below explanation of this property.
hibernate.connection.driver_class	-	Driver class to use by hibernate, if not specified the framework will auto asign one
hibernate.dialect	-	Dialect to use by hibernate, if not specified the framework will auto asign one. Check https://docs.jboss.org/hibernate/orm/4.3/manual/en-US/html/ch03.html#configuration-optional-dialects for a complete list of available dialects
hibernate.connection.provider_class	-	Set to org.hibernate.connection.C3P0ConnectionProvider to use C3P0 connection pool (recommended for production)
hibernate.c3p0.min_size	-	Min connection pool size
hibernate.c3p0.max_size	-	Max connection pool size
default.charset.resultset	UTF-8	Result set from DB converted to charset character encoding

Standard Query

If no custom query is set, SELECT <columns.to.select> FROM <table> will be executed each run.query.delay milliseconds configured

Custom Query

A custom query is supported to bring the possibility of using the entire SQL language. This is powerful, but risky, be careful with the custom queries used.

To avoid row export repetitions use the $@$ special character in WHERE clause, to incrementaly export not processed rows and the new ones inserted.

IMPORTANT: For proper operation of Custom Query ensure that incremental field will be returned in the first position of the Query result.

Example:

agent.sources.sql-source.custom.query = SELECT incrementalField,field2 FROM table1 WHERE incrementalField > $@$ 

Configuration example

# For each one of the sources, the type is defined
agent.sources.sqlSource.type = org.keedio.flume.source.SQLSource

agent.sources.sqlSource.hibernate.connection.url = jdbc:db2://192.168.56.70:50000/sample

# Hibernate Database connection properties
agent.sources.sqlSource.hibernate.connection.user = db2inst1
agent.sources.sqlSource.hibernate.connection.password = db2inst1
agent.sources.sqlSource.hibernate.connection.autocommit = true
agent.sources.sqlSource.hibernate.dialect = org.hibernate.dialect.DB2Dialect
agent.sources.sqlSource.hibernate.connection.driver_class = com.ibm.db2.jcc.DB2Driver

#agent.sources.sqlSource.table = employee1

# Columns to import to kafka (default * import entire row)
#agent.sources.sqlSource.columns.to.select = *

# Query delay, each configured milisecond the query will be sent
agent.sources.sqlSource.run.query.delay=10000

# Status file is used to save last readed row
agent.sources.sqlSource.status.file.path = /var/log/flume
agent.sources.sqlSource.status.file.name = sqlSource.status

# Custom query
agent.sources.sqlSource.start.from = 19700101000000000000
agent.sources.sqlSource.custom.query = SELECT * FROM (select DECIMAL(test) * 1000000 AS INCREMENTAL, EMPLOYEE1.* from employee1 UNION select DECIMAL(test) * 1000000 AS INCREMENTAL, EMPLOYEE2.* from employee2) WHERE INCREMENTAL > $@$ ORDER BY INCREMENTAL ASC

agent.sources.sqlSource.batch.size = 1000
agent.sources.sqlSource.max.rows = 1000
agent.sources.sqlSource.delimiter.entry = |

agent.sources.sqlSource.hibernate.connection.provider_class = org.hibernate.connection.C3P0ConnectionProvider
agent.sources.sqlSource.hibernate.c3p0.min_size=1
agent.sources.sqlSource.hibernate.c3p0.max_size=10

# The channel can be defined as follows.
agent.sources.sqlSource.channels = memoryChannel

Known Issues

An issue with Java SQL Types and Hibernate Types could appear Using SQL Server databases and SQL Server Dialect coming with Hibernate.

Something like:

org.hibernate.MappingException: No Dialect mapping for JDBC type: -15

Use org.keedio.flume.source.SQLServerCustomDialect in flume configuration file to solve this problem.

Special thanks

I used flume-ng-kafka to guide me (https://github.com/baniuyao/flume-ng-kafka-source.git). Thanks to Frank Yao.

Version History

• Version 1.5.1 added charset encoding for result set is now configurable.
• Stable version is 1.5.0 (compatible with Apache Flume 1.8.0)
• Previous stable version is 1.4.3 (compatible with Apache Flume prior to 1.7.0)

问题， 如果在导入数据到hdfs, 需要将hadoop相关的jar包放到flume lib目录下