841      阿里云  技术社区[云栖]

Flume MaxCompute Sink插件__数据入云_数据集成-阿里云

Apache Flume是一个分布式的、可靠的、可用的系统，可用于从不同的数据源中高效地收集、聚合和移动海量日志数据到集中式数据存储系统。

ODPS Sink是基于ODPS DataHub Service开发的Flume插件，可以将Flume的Event数据导入到ODPS中。插件兼容Flume的原有功能特性，支持ODPS表自定义分区、且可以自动创建分区。

二、环境要求

1、JDK(1.6以上，推荐1.7)

2、Flume-NG 1.x

三、插件部署

1、下载ODPS Sink插件并解压：aliyun-odps-flume-plugin

2、部署ODPS Sink插件：将文件夹odps_sink移动到Apache Flume安装目录下：

$ mkdir {YOUR_APACHE_FLUME_DIR}/plugins.d
$mv  odps_sink/  { YOUR_APACHE_FLUME_DIR }/plugins.d/

移动后，核验ODPS Sink插件是否已经在相应目录：

$ ls { YOUR_APACHE_FLUME_DIR}/plugins.d
odps_sink

部署完成后，只需要在Flume的配置文件中将sink的type字段配置为:

com.aliyun.odps.flume.sink.OdpsSink

即可使用

四、配置示例

例：将日志文件中的结构化数据进行解析，并上传到ODPS表中

需要上传的日志文件格式如下（每行为一条记录，字段之间逗号分隔）:

#test_basic.log
some,log,line1
some,log,line2
...

第一步、在ODPS 的 project创建ODPS Datahub表

建表语句如下所示：

CREATE TABLE hub_table_basic (col1 STRING, col2 STRING) 
    PARTITIONED BY (pt STRING)
    INTO 1 SHARDS
    HUBLIFECYCLE 1;

第二步、创建Flume作业配置文件：

在Flume安装目录的conf/文件夹下创建名为odps_basic.conf的文件，并输入内容如下：

# odps_basic.conf
# A single-node Flume configuration for ODPS
# Name the components on this agent
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# Describe/configure the source
a1.sources.r1.type = exec
a1.sources.r1.command = cat {YOUR_LOG_DIRECTORY}/test_basic.log
# Describe the sink
a1.sinks.k1.type = com.aliyun.odps.flume.sink.OdpsSink
a1.sinks.k1.accessID = {YOUR_ALIYUN_ODPS_ACCESS_ID}
a1.sinks.k1.accessKey = {YOUR_ALIYUN_ODPS_ACCESS_KEY}
a1.sinks.k1.odps.endPoint = https://service.odps.aliyun.com/api
a1.sinks.k1.odps.datahub.endPoint = https://dh.odps.aliyun.com
a1.sinks.k1.odps.project =  {YOUR_ALIYUN_ODPS_PROJECT}
a1.sinks.k1.odps.table = hub_table_basic
a1.sinks.k1.odps.partition = 20150814
a1.sinks.k1.batchSize = 100
a1.sinks.k1.serializer = DELIMITED
a1.sinks.k1.serializer.delimiter = ,
a1.sinks.k1.serializer.fieldnames = col1,,col2
a1.sinks.k1.serializer.charset = UTF-8
a1.sinks.k1.shard.number = 1
a1.sinks.k1.shard.maxTimeOut = 60
a1.sinks.k1.autoCreatePartition = true
# Use a channel which buffers events in memory
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 1000
# Bind the source and sink to the channel
a1.sources.r1.channels = c1
a1.sinks.k1.channel = c1

$ cd { YOUR_APACHE_FLUME_DIR}
$ bin/flume-ng agent -n a1 -c conf -f conf/odps_basic.conf -Dflume.root.logger=INFO,console

...
Write success. Event count: 2
...