flume学习笔记

最新推荐文章于 2022-12-01 10:56:20 发布
最新推荐文章于 2022-12-01 10:56:20 发布
阅读量556 收藏
点赞数
分类专栏： 云计算 flume
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/shandadadada/article/details/50879202
版权

１.Flume介绍及设计思想
一．介绍
Flume是一个分布式、可靠、和高可用的海量日志采集、聚合和传输的系统。

目的：把不同数据源的数据，汇集到一个集中存储中心。

版本： 0.9.x(OG) VS 1.x(NG)　还没了解OG和NG的区别

特点：分布式／可靠及高可用性／实时性／扩展性

二．设计思想

模块化设计：内部按职能划分Source、Channel、Sink等模块
组合式设计：根据不同场景选择适合的Source、Channel、Sink进行组合
数据流模型：支持复杂的数据流（fan-in、fan-out），构建自己的拓扑
扩展性/插件式设计：可以根据自己业务需要来定制实现某些组件
可靠保障
数据存储于各agent的channel中，用事务来保证agent之间的数据传递
flume提供基于文件的channel

三．基本组件
这里写图片描述
Agent：一个Flume JVM进程，可包含多个Source，Sink，Channel
Source：采集数据，写入Channel
Sink：读取Channel数据，持久化
Channel：连接Source及Sink
Event：数据的传递形式，包含header及body两部分
拓扑：Flume传输的架构，通过串联Agent及三大组件形成
这里写图片描述

四：常用组件
Source：Avro、Thrift、Syslog、Netcat、Exec、Http
Sink：HDFS、Kafka、Avro、File、HBase
Channel：Memory、File、Kafka、Spillable-Memory

五：使用场景

什么时候使用fan-out？什么时候使用fan-in?

fan-out:比如我们接受tomcat日志，这时候日志可能会打到kafka／es等不同channel处理
fan-in:比如我们一台机器上有两台tomcat，这时候两个source都写入到同一个channel.

２.安装
一．下载地址
http://www.apache.org/dyn/closer.lua/flume/1.6.0/apache-flume-1.6.0-bin.tar.gz

JDK
1.6 or later (Java 1.7 Recommended)

二．配置
配置文件目录：apache-flume-1.6.0-bin/conf
flume-env.sh：环境变量（JDK、Classpath）
flume-conf.properties：Agent配置

配置样例：
# list the sources, sinks and channels for the agent
<Agent>.sources = <Source>
<Agent>.sinks = <Sink>
<Agent>.channels = <Channel1> <Channel2>

# set channel for source
<Agent>.sources.<Source>.channels = <Channel1> <Channel2> ...

# set channel for sink
<Agent>.sinks.<Sink>.channel = <Channel1>

启动命令：

bin/flume-ng agent --conf ./conf/ -f conf/flume-conf.properties 
-Dflume.root.logger=DEBUG,console -n <Agent>

３.实践

1.example01：通过tcp接收数据，然后打印出来

# ============== test read telnet ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example01.conf -Dflume.root.logger=INFO,console -n a1

a1.sources = r1
a1.channels = c1
a1.sinks = k1

a1.sources.r1.type = netcat
a1.sources.r1.channels = c1
a1.sources.r1.bind = 0.0.0.0
a1.sources.r1.port = 4140

a1.channels.c1.type = memory
a1.channels.c1.capacity = 100

a1.sinks.k1.type = logger
a1.sinks.k1.channel = c1

telnet:

pc:~/software/flume/apache-flume-1.6.0-bin$ telnet localhost 4140
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
dad
OK
dafdfasdf
OK

kafka输出：

2016-03-13 18:02:10,999 (SinkRunner-PollingRunner-DefaultSinkProcessor) [INFO - org.apache.flume.sink.LoggerSink.process(LoggerSink.java:94)] Event: { headers:{} body: 64 61 64 0D                                     dad. }
2016-03-13 18:02:16,200 (SinkRunner-PollingRunner-DefaultSinkProcessor) [INFO - org.apache.flume.sink.LoggerSink.process(LoggerSink.java:94)] Event: { headers:{} body: 64 61 66 64 66 61 73 64 66 0D                   dafdfasdf. }

2.example02：tail一个文件，然后写入Kafka

# ============== test tail file ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example02.conf -Dflume.root.logger=INFO,console -n a1

a1.sources = r1
a1.channels = c1
a1.sinks = k1

a1.sources.r1.type = exec
a1.sources.r1.command = tail -F /tmp/flume.test
a1.sources.r1.channels = c1

a1.channels.c1.type = memory
a1.channels.c1.capacity = 100

a1.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
a1.sinks.k1.channel = c1
a1.sinks.k1.topic = test.flume.topic
a1.sinks.k1.brokerList = 192.168.6.13:9092
a1.sinks.k1.requiredAcks = 0
a1.sinks.k1.batchSize = 2000
a1.sinks.k1.kafka.producer.type = async
a1.sinks.k1.kafka.batch.num.messages = 1000
echo 2323sd >> flume.test
bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic test.flume.topic --from-beginning
xueguolin@xueguolin-pc:~/software/kafka/kafka_2.10-0.8.2.0$ [2016-05-11 22:38:17,557] INFO Closing socket connection to /127.0.0.1. (kafka.network.Processor)
bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic test.flume.topic --from-beginning
[2016-05-11 22:38:24,397] INFO Closing socket connection to /127.0.0.1. (kafka.network.Processor)
23
[2016-05-11 22:41:34,727] INFO Closing socket connection to /127.0.0.1. (kafka.network.Processor)
qw
we

sd
2323sd
2323sd
2323sd
2323sd
2323sd
adad

adad2323

3.example03:tail两个文件,在聚集端打印出来
agent:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example03-agent.conf -Dflume.root.logger=INFO,console -n flumeAgent

flumeAgent.sources = r1 r2
flumeAgent.channels = c1
flumeAgent.sinks = k1

flumeAgent.sources.r1.type = exec
flumeAgent.sources.r1.command = tail -F /tmp/flume.test1
flumeAgent.sources.r1.channels = c1

flumeAgent.sources.r2.type = exec
flumeAgent.sources.r2.command = tail -F /tmp/flume.test2
flumeAgent.sources.r2.channels = c1

flumeAgent.channels.c1.type = memory
flumeAgent.channels.c1.capacity = 100

flumeAgent.sinks.k1.type = avro
flumeAgent.sinks.k1.channel = c1
flumeAgent.sinks.k1.hostname = 127.0.0.1
flumeAgent.sinks.k1.port = 4141

server:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example03-server.conf -Dflume.root.logger=INFO,console -n flumeServer

flumeServer.sources = r1
flumeServer.channels = c1
flumeServer.sinks = k1

flumeServer.sources.r1.type = avro
flumeServer.sources.r1.channels = c1
flumeServer.sources.r1.bind = 0.0.0.0
flumeServer.sources.r1.port = 4141

flumeServer.channels.c1.type = memory
flumeServer.channels.c1.capacity = 100

flumeServer.sinks.k1.type = logger
flumeServer.sinks.k1.channel = c1

xueguolin@xueguolin-pc:/tmp$ echo 12 >> flume.test1
xueguolin@xueguolin-pc:/tmp$ echo 1223 >> flume.test2
2016-05-15 14:47:48,273 (SinkRunner-PollingRunner-DefaultSinkProcessor) [INFO - org.apache.flume.sink.LoggerSink.process(LoggerSink.java:94)] Event: { headers:{} body: 31 32                                           12 }
2016-05-15 14:48:03,274 (SinkRunner-PollingRunner-DefaultSinkProcessor) [INFO - org.apache.flume.sink.LoggerSink.process(LoggerSink.java:94)] Event: { headers:{} body: 31 32 32 33                                     1223 }

问题1：如何区分example03的两个日志？
interceptor：可在Source把Event写入Channel之前，对Event做一些操作
selector：channel选择器

添加interceptor之后的配置：
flumeAgent.sources.r1.interceptors = i1
flumeAgent.sources.r1.interceptors.i1.type = static
flumeAgent.sources.r1.interceptors.i1.key = whoami
flumeAgent.sources.r1.interceptors.i1.value = log1


flumeAgent.sources.r2.interceptors = i1
flumeAgent.sources.r2.interceptors.i1.type = static
flumeAgent.sources.r2.interceptors.i1.key = whoami
flumeAgent.sources.r2.interceptors.i1.value = log2
添加多路选择之后的配置：
selector.type
replicating：所有Channel都写入数据，默认值
multiplexing：根据规则选择Channel写数据

flumeServer.sources.r1.channels = c1 c2
flumeServer.sources.r1.selector.type = multiplexing
flumeServer.sources.r1.selector.header = whoami
flumeServer.sources.r1.selector.mapping.log1 = c1
flumeServer.sources.r1.selector.mapping.log2 = c2

例子:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example03-q1-agent.conf -Dflume.root.logger=INFO,console -n flumeAgent

flumeAgent.sources = r1 r2
flumeAgent.channels = c1
flumeAgent.sinks = k1

flumeAgent.sources.r1.type = exec
flumeAgent.sources.r1.command = tail -F /tmp/flume.test1
flumeAgent.sources.r1.channels = c1
flumeAgent.sources.r1.interceptors = i1
flumeAgent.sources.r1.interceptors.i1.type = static
flumeAgent.sources.r1.interceptors.i1.key = whoami
flumeAgent.sources.r1.interceptors.i1.value = log1

flumeAgent.sources.r2.type = exec
flumeAgent.sources.r2.command = tail -F /tmp/flume.test2
flumeAgent.sources.r2.channels = c1
flumeAgent.sources.r2.interceptors = i1
flumeAgent.sources.r2.interceptors.i1.type = static
flumeAgent.sources.r2.interceptors.i1.key = whoami
flumeAgent.sources.r2.interceptors.i1.value = log2

flumeAgent.channels.c1.type = memory
flumeAgent.channels.c1.capacity = 100

flumeAgent.sinks.k1.type = avro
flumeAgent.sinks.k1.channel = c1
flumeAgent.sinks.k1.hostname = 127.0.0.1
flumeAgent.sinks.k1.port = 4141
# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example03-q1-server.conf -Dflume.root.logger=INFO,console -n flumeServer

flumeServer.sources = r1
flumeServer.channels = c1 c2
flumeServer.sinks = k1 k2

flumeServer.sources.r1.type = avro
flumeServer.sources.r1.channels = c1 c2
flumeServer.sources.r1.bind = 0.0.0.0
flumeServer.sources.r1.port = 4141
flumeServer.sources.r1.selector.type = multiplexing
flumeServer.sources.r1.selector.header = whoami
flumeServer.sources.r1.selector.mapping.log1 = c1
flumeServer.sources.r1.selector.mapping.log2 = c2

#==============================================
flumeServer.channels.c1.type = memory
flumeServer.channels.c1.capacity = 100

flumeServer.channels.c2.type = memory
flumeServer.channels.c2.capacity = 100

#==============================================
flumeServer.sinks.k1.type = logger
flumeServer.sinks.k1.channel = c1

flumeServer.sinks.k2.type = logger
flumeServer.sinks.k2.channel = c2
xueguolin@xueguolin-pc:/tmp$ echo 1223 >> flume.test2
xueguolin@xueguolin-pc:/tmp$ echo 1333223 >> flume.test2
xueguolin@xueguolin-pc:/tmp$ echo 12kk >> flume.test1

2016-05-15 14:54:07,313 (SinkRunner-PollingRunner-DefaultSinkProcessor) [INFO - org.apache.flume.sink.LoggerSink.process(LoggerSink.java:94)] Event: { headers:{whoami=log2} body: 31 32 32 33                                     1223 }
2016-05-15 14:54:38,502 (SinkRunner-PollingRunner-DefaultSinkProcessor) [INFO - org.apache.flume.sink.LoggerSink.process(LoggerSink.java:94)] Event: { headers:{whoami=log2} body: 31 33 33 33 32 32 33                            1333223 }
2016-05-15 14:55:01,318 (SinkRunner-PollingRunner-DefaultSinkProcessor) [INFO - org.apache.flume.sink.LoggerSink.process(LoggerSink.java:94)] Event: { headers:{whoami=log1} body: 31 32 6B 6B                                     12kk }

问题2：example03只有一个Server，挂了怎么办？
sinkgroup：把多个sink捆绑成一个

processor.type
default：默认值，单sink
failover：容错处理（异常sink进入cd阶段）
load_balance：负载均衡（也具有容错功能，但会挨个尝试sink）

添加sinkgroup：
flumeAgent.sinkgroups = g1
flumeAgent.sinkgroups.g1.sinks = k1 k2
flumeAgent.sinkgroups.g1.processor.type = failover

例子:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example03-q2-agent.conf -Dflume.root.logger=INFO,console -n flumeAgent

flumeAgent.sources = r1 r2
flumeAgent.channels = c1
flumeAgent.sinks = k1 k2

flumeAgent.sources.r1.type = exec
flumeAgent.sources.r1.command = tail -F /tmp/flume.test1
flumeAgent.sources.r1.channels = c1

flumeAgent.sources.r2.type = exec
flumeAgent.sources.r2.command = tail -F /tmp/flume.test2
flumeAgent.sources.r2.channels = c1

flumeAgent.channels.c1.type = memory
flumeAgent.channels.c1.capacity = 100


flumeAgent.sinkgroups = g1
flumeAgent.sinkgroups.g1.sinks = k1 k2
flumeAgent.sinkgroups.g1.processor.type = failover

flumeAgent.sinks.k1.type = avro
flumeAgent.sinks.k1.channel = c1
flumeAgent.sinks.k1.hostname = 127.0.0.1
flumeAgent.sinks.k1.port = 4141
flumeAgent.sinks.k1.batch-size = 1

flumeAgent.sinks.k2.type = avro
flumeAgent.sinks.k2.channel = c1
flumeAgent.sinks.k2.hostname = 127.0.0.1
flumeAgent.sinks.k2.port = 4142
flumeAgent.sinks.k2.batch-size = 1

server1:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example03-q2-server1.conf -Dflume.root.logger=INFO,console -n flumeServer

flumeServer.sources = r1
flumeServer.channels = c1
flumeServer.sinks = k1

flumeServer.sources.r1.type = avro
flumeServer.sources.r1.channels = c1
flumeServer.sources.r1.bind = 0.0.0.0
flumeServer.sources.r1.port = 4141

flumeServer.channels.c1.type = memory
flumeServer.channels.c1.capacity = 100

flumeServer.sinks.k1.type = logger
flumeServer.sinks.k1.channel = c1

server2:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example03-q2-server2.conf -Dflume.root.logger=INFO,console -n flumeServer

flumeServer.sources = r1
flumeServer.channels = c1
flumeServer.sinks = k1

flumeServer.sources.r1.type = avro
flumeServer.sources.r1.channels = c1
flumeServer.sources.r1.bind = 0.0.0.0
flumeServer.sources.r1.port = 4142

flumeServer.channels.c1.type = memory
flumeServer.channels.c1.capacity = 100

flumeServer.sinks.k1.type = logger
flumeServer.sinks.k1.channel = c1

两个server发给kafka:

agent:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example04-kafka-agent.conf -Dflume.root.logger=INFO,console -n flumeAgent

flumeAgent.sources = r1 r2
flumeAgent.channels = c1
flumeAgent.sinks = k1 k2

flumeAgent.sources.r1.type = exec
flumeAgent.sources.r1.command = tail -F /tmp/flume.test1
flumeAgent.sources.r1.channels = c1

flumeAgent.sources.r2.type = exec
flumeAgent.sources.r2.command = tail -F /tmp/flume.test2
flumeAgent.sources.r2.channels = c1

flumeAgent.channels.c1.type = memory
flumeAgent.channels.c1.capacity = 100


flumeAgent.sinkgroups = g1
flumeAgent.sinkgroups.g1.sinks = k1 k2
flumeAgent.sinkgroups.g1.processor.type = failover

flumeAgent.sinks.k1.type = avro
flumeAgent.sinks.k1.channel = c1
flumeAgent.sinks.k1.hostname = 127.0.0.1
flumeAgent.sinks.k1.port = 4141
flumeAgent.sinks.k1.batch-size = 1

flumeAgent.sinks.k2.type = avro
flumeAgent.sinks.k2.channel = c1
flumeAgent.sinks.k2.hostname = 127.0.0.1
flumeAgent.sinks.k2.port = 4142
flumeAgent.sinks.k2.batch-size = 1

server1:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example04-kafka-server1.conf -Dflume.root.logger=INFO,console -n flumeServer

flumeServer.sources = r1
flumeServer.channels = c1
flumeServer.sinks = k1

flumeServer.sources.r1.type = avro
flumeServer.sources.r1.channels = c1
flumeServer.sources.r1.bind = 0.0.0.0
flumeServer.sources.r1.port = 4141

flumeServer.channels.c1.type = memory
flumeServer.channels.c1.capacity = 100

flumeServer.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
flumeServer.sinks.k1.channel = c1
flumeServer.sinks.k1.topic = test.flume.topic
flumeServer.sinks.k1.brokerList = 127.0.0.1:9092
flumeServer.sinks.k1.requiredAcks = 0
flumeServer.sinks.k1.batchSize = 2000
flumeServer.sinks.k1.kafka.producer.type = sync
flumeServer.sinks.k1.kafka.batch.num.messages = 1000

server2:

# ============== test write kafka ========================================
# bin/flume-ng agent --conf ./conf/ -f conf/example04-kafka-server2.conf -Dflume.root.logger=INFO,console -n flumeServer

flumeServer.sources = r1
flumeServer.channels = c1
flumeServer.sinks = k1

flumeServer.sources.r1.type = avro
flumeServer.sources.r1.channels = c1
flumeServer.sources.r1.bind = 0.0.0.0
flumeServer.sources.r1.port = 4142

flumeServer.channels.c1.type = memory
flumeServer.channels.c1.capacity = 100

flumeServer.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
flumeServer.sinks.k1.channel = c1
flumeServer.sinks.k1.topic = test.flume.topic
flumeServer.sinks.k1.brokerList = 127.0.0.1:9092
flumeServer.sinks.k1.requiredAcks = 0
flumeServer.sinks.k1.batchSize = 2000
flumeServer.sinks.k1.kafka.producer.type = sync
flumeServer.sinks.k1.kafka.batch.num.messages = 1000

kafka:

xueguolin@xueguolin-pc:/tmp$ echo 1333223ddd >> flume.test1
xueguolin@xueguolin-pc:/tmp$ echo 1333223ddd >> flume.test1
xueguolin@xueguolin-pc:/tmp$ echo 1333223ddd >> flume.test1
xueguolin@xueguolin-pc:/tmp$ echo 1333222223ddkafafad >> flume.test1



xueguolin@xueguolin-pc:~/software/kafka/kafka_2.10-0.8.2.0$ bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic test.flume.topic --from-beginning
12



12
1333223ddd
1333223ddd
1333222223ddd
1333222223ddkafafad
shandadadada
关注
专栏目录
Flume笔记
Yuexp0825的博客
04-13 148
Flume 概述: 特点：高可用的，高可靠的，分布式的海量日志采集、聚合和传输的系统。 选Flume的原因：FLume是流式传输实时处理的。 数据的传输过程：没有将用户产生的日志直接传输给大数据平台,而是传给日志服务器,可以按自己的需求方式去日志服务器获取。 不希望以你的方式来传输而时希望以我的方式去拿。 日志：<1>框架工作时候产生日志;<2>用户的操作行为被后台服务器收集后产生的用户行为日志。 Log4j：通过日志（log4j）的方式打印日志 日志的级别：...
Flume学习的一些笔记
J1NX1N的博客
08-29 182
1、Source、Channel、sink Source        目前Flume中Source基本是Taildir Source，它可以实现断点续传，以及监控多目录，在Flume1.6以前需要自定义Source记录每次读取的文件位置，实现断点续传。Exec Source虽然可以实时搜集数据集，但在Flume不运行或者Shell出错时，数据会丢失。而SpoolingDirectory source能监控目录，但是不能断点续传，所以推荐
Flume学习笔记
ZeroTeam_麒麟
04-08 1525
Flume学习笔记和实践分享一
flume笔记
weixin_45764675的博客
02-06 216
Flume是什么? Flume是⼀种分布式，可靠且可⽤的服务，⽤于有效地收集，聚合和移动⼤量⽇志数据。Flume构建在⽇志流之上⼀个简单灵活的架构。它具有可靠的可靠性机制和许多故障转移和恢复机制，具有强⼤的容错性。使⽤Flume这套架构实现对⽇志流数据的实时在线分析。Flume⽀持在⽇志系统中定制各类数据发送⽅，⽤于收集数据；同时，Flume提供对数据进⾏简单处理，并写到各种数据接受⽅（可定制）的...
Flume 笔记
qq_36864672的博客
11-28 463
$>flume-ng avro-client -H localhost -p 4141 -F /xxx(文件)  上面出错改成这个                   $>flume-ng agent -z s101:2181,s102:2181 -p /flume --name a1 -Dflume.root.logger=INFO,console 通过这个连接： $>flume
flume学习笔记.pptx
09-23
flume学习笔记 ===亲测===== 笔记中所有demo都已由本人实现
flume学习笔记更新版20190927.pptx
09-27
更新flume笔记文档 其中包括flume安装，flume监控文件、目录等并保存数据 还包含flume启动脚本
flume学习笔记（一）：安装flume
最新发布
weixin_41765352的博客
12-01 456
这是本人近期学习flume的笔记，供和我一样投身大数据技术海洋的小白有个参考。本笔记使用的flume版本：flume1.9。
Flume 学习笔记（一）使用Flume收集netcat端口通信的信息到本地文件
希望每天都能进步一点点
03-19 1478
若发现文章有误，敬请指教，感谢 文章目录运行环境一、准备 netcat 工具二、认识 Flume2.1 Flume 概念2.1 Flume 架构三、使用 Flume 收集端口的信息3.1 开启netcat3.2 创建配置文件3.3 启动flume 运行环境 Vmware 15 Flume1.9 CentOS7 操作系统 一、准备 netcat 工具 参考资料 Netcat 是一款简单的Unix工具，使用UDP和TCP协议。 它是一个可靠的容易被其他程序所启用的后台操作工具，同时它也被用作网络的.
Flume学习笔记（一）Flume 组成架构
an_vana的博客
08-08 2760
Flume学习笔记（一）Flume 组成架构 前言 ​ 本文主要记录我在学习 Flume 过程中的一些知识的整理与记录，预计会做成一个系列来梳理一下 Flume 中的知识，本篇的主要内容为 Flume 的组成架构，文中如有疏漏与不足欢迎指正！ 一、Flume 是什么 ​ Flume 是 Cloudera 提供的一个高可用的，高可靠的，分布式的海量日志采集、聚合和传输的系统。Flume 基于流式架构，灵活简单。其可以不断的监听数据源来将新产生的数据搬迁至目标，如将新产生的log 日志数据写入 HDFS。 二、
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值