原理：通过采集Agent获取DB数据，采集数据存储在Hippo，TDSort消费Hippo数据，存储到对应目的DB
如果接口名不符合命名规范，则数据无法落地的期望的目的。TDSort消费时会抛弃该数据
Hippo创建，数据接入创建后，需要所属项目管理员进行审批，只有审批通过后，才能正常使用。审批成功后，数据接入从草稿变成启动状态。
data.hub.manager.ip={DataHub Server对应的IP地址}
data.hub.manager.port={DataHub Server对应的heartbeat.port配置，默认:9292}
data.hub.agent.name=agent-db-test
data.hub.name=db2dbTest #对应接入名称
data.hub.source=MySQL #数据类型
data.hub.target=DB #落地的数据类型
log4j.logger.metrics=INFO, METRICS
log4j.appender.METRICS=org.apache.log4j.rolling.RollingFileAppender
log4j.appender.METRICS.rollingPolicy=org.apache.log4j.rolling.TimeBasedRollingPolicy
log4j.appender.METRICS.rollingPolicy.ActiveFileName=${flume.log.dir}/metrics.log
log4j.appender.METRICS.rollingPolicy.FileNamePattern=${flume.log.dir}/metrics.log.%d{yyyy-MM-dd}
log4j.appender.METRICS.layout=org.apache.log4j.PatternLayout
log4j.appender.METRICS.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss} %m%n
agent1.sources = sqlSource
agent1.channels = ch1
##################### HippoChannel #####################
agent1.channels.ch1.type = com.tencent.tbds.datahub.agent.flume.channel.hippo.HippoChannel
agent1.channels.ch1.hippo.topic = db_topic_01 #接入对应的HippoTopic名称
agent1.channels.ch1.hippo.controllerIpList = 10.0.0.2:8066 #HIPPO Controller对应IP和跟broker的心跳端口，默认：8066
agent1.channels.ch1.hippo.producerGroup = group_db_001 #生产hippo数据的group组
agent1.channels.ch1.hippo.producer.secretId = {secureId在Hippo topic的详情页面查看}
agent1.channels.ch1.hippo.producer.secretKey = {secureKey在Hippo topic的详情页面查看}
agent1.channels.ch1.hippo.producer.sendAsync.fqueue.path = /data/fqueue
##################### 全量数据采集 #####################
agent1.sources.sqlSource.channels = ch1
agent1.sources.sqlSource.type = com.tencent.tbds.tdbank.flume.source.sql.SQLSource
agent1.sources.sqlSource.hibernate.connection.url = jdbc:mysql://10.0.0.1:3306/tdbank #源数据库连接信息
agent1.sources.sqlSource.hibernate.connection.user = root
agent1.sources.sqlSource.hibernate.connection.password = 123456
agent1.sources.sqlSource.hibernate.connection.isolation = 2
agent1.sources.sqlSource.hibernate.dialect = org.hibernate.dialect.MySQL5Dialect
agent1.sources.sqlSource.hibernate.connection.driver_class = com.mysql.jdbc.Driver
agent1.sources.sqlSource.database = tdbank #采集数据源DB
agent1.sources.sqlSource.tables = student,teacher #采集数据源表，多个表逗号分隔
## 增加以下配置可使用索引字段（假设为id）提升查询性能
## 另外也可以使用自定义SQL
## agent1.sources.sqlSource.tables.tbl_poc1.whereClause = WHERE id > :index ORDER BY id
## agent1.sources.sqlSource.tables.tbl_poc1.column.name = id
agent1.sources.sqlSource.recoveryMode = true #true代表数据采集完成后自动退出
agent1.sources.sqlSource.status.filePath = /data/sqlStatus
agent1.sources.sqlSource.hibernate.connection.provider_class = org.hibernate.connection.C3P0ConnectionProvider
agent1.sources.sqlSource.hibernate.c3p0.min_size=1
agent1.sources.sqlSource.hibernate.c3p0.max_size=10
[agent]
;agent主机地址，非必填项；默认使用本机ip
hostName=10.0.0.2
;agent监听的rpc端口，必填项；
port=8003 #Tmetrics Agents对应rpc端口，默认8003
masterAddressList=10.0.0.4:9000 #Tmetrics Master的Ip和Master对应rpc端口，默认9000
bdbFilePath=/data/metric-2.0/bdb
metricSpliter=#
;配置指标文件路径
metricFiles=/data/flume/logs
;配置单条指标的模式
metricPartern1=hippo@PRODUCER#topic#group#clientIp#time#agentCnt#agentPkg#agentSize@true@4#yyyyMMddHHmm@70@5#6#7
metricPartern2=bus@TDBus#topic#interface#group#clientIp#brokerIp#time#agentPkg#agentSize@true@6#yyyyMMddHHmm@120000@7#8
metricSuffixPartern=[1-9][0-9]{3}-[0-1][0-9]-[0-3][0-9]
metricPrefixDateFormater=yyyy-MM-dd HH:mm:ss
metricAggrThreadNum=2
metricSendThreadNum=2
maxCntOnePkg=200
partAggrCacheSize=5000
needCompress=true
queueFlushInterval=10000
heartbeatPeriodInMills=5000
isShouldPrint=true
;FQueue
queueCleanInterval=10
queueDataTTL=20
failoverQueuePath=/data/metric-2.0/failover
cd ${FLUME_PATH}
nohup bin/flume-ng agent -Dlog4j.configuration=conf/log4j.properties --classpath /{DataHub_PATH}/conf:/{DataHub_PATH}/lib/* --conf conf/ -f conf/flume.conf -n agent1 > stdout 2>&1 &
cd ${TMetric_PATH}
bin/agent.sh start
