Introduction
This framework is based on the second package Netty Http interface service performance, increases the functionality of the http request routing, and simplifies the operation, the received object is a log simply treated bolus to Kafka, serving ECHINFO buried data show business, daily processing 1 billion during the Spring Festival + features: simple and efficient (the most common environmental pressure measuring machine can reach a maximum QPS 2.5w / s) project address: github.com/eqxiu/log-s...
design
rely
- Netty4
- logback kafka appender
- IPIP
Quick Start
1, the compiler
git clone https://github.com/yanchaoguo/log-server.git
mvn assembly:assembly
复制代码2, configure the compiled project will put / data / work / log_server, the directory structure is as follows:
----log_server
|
----bin #依赖的jar包
|
----classes #编译后生成的classes目录
|
----bin #启动脚本
|
----logs #日志文件
复制代码cd / data / work / log_server / bin ls.sh script editor, configure the port number and address kafka
port=9001
kafka=hadoop006:9092,hadoop007:9092,hadoop008:9092
复制代码3, start
./ls.sh start
复制代码Rapid Development
1, catalog descriptions
2, in the new action-based directory service processing, such as to create and implement FastPushAction doAction Action interface method; such log not processed, not much business logic, responsible for sending the received logs to Kafka, can be used as a reference demo , to achieve the following specific
/**
* /fast_push为该业务逻辑的请求路径 ，如http://localhost:port/log-server/fast_push
*/
@Route(value = "/fast_push")
public class FastPushAction implements Action {
private static final Logger logger = LoggerFactory.getLogger(PushLogAction.class);
@Override
public void doAction(Request request, Response response) {
String logs = request.getContent(); //从body中获取日志内容
//从参数中获取 logger 名称，为空则取默认值
String loger = Utils.isNulDefault(request.getParam("loger"),LogConfigManager.trashTopic);
int responseStatus = CodeManager.RESPONSE_CODE_NORMAL;
//如果body中没有内容 尝试从url参数中获取
if (StrUtil.isEmpty(logs) && request.getParams().size()>0) {
logs = Utils.toJson(request.getParams());
}
if (StrUtil.isEmpty(logs)) {
response.setContent("<h2>not find log content</h2>");
responseStatus = CodeManager.RESPONSE_CODE_NOT_CONTENT;
logger.warn("not find log content");
}else {
//将日志推送到kafka，其中 logger 的名称要和logback.xml中的配置一致
KafkaLoggerFactory.getLogger(loger).info(logs);
}
// 设置返回信息：
response.setStatus(responseStatus);
response.setHeader("Access-Control-Allow-Origin", "*");
response.setHeader(CookieHeaderNames.EXPIRES , -1);
// 返回：
response.send();
}
}
复制代码3, configuration logback.xml
<?xml version="1.0" encoding="UTF-8" ?>
<configuration>
<!--trash-->
<logger name="trash" level="INFO" additivity="false">
<appender-ref ref="trash_kafka"/>
</logger>
<appender name="trash_kafka" class="com.github.danielwegener.logback.kafka.KafkaAppender">
<encoder class="com.github.danielwegener.logback.kafka.encoding.LayoutKafkaMessageEncoder">
<layout class="ch.qos.logback.classic.PatternLayout">
<pattern>%msg</pattern>
</layout>
<charset>UTF-8</charset>
</encoder>
<topic>trash</topic>
<keyingStrategy class="com.github.danielwegener.logback.kafka.keying.RoundRobinKeyingStrategy" />
<deliveryStrategy class="com.github.danielwegener.logback.kafka.delivery.AsynchronousDeliveryStrategy" />
<producerConfig>bootstrap.servers=${kafka.servers}
</producerConfig>
<producerConfig>linger.ms=1000</producerConfig>
<producerConfig>compression.type=none</producerConfig>
<producerConfig>buffer.memory=1073741842</producerConfig>
<producerConfig>acks=1</producerConfig>
</appender>
</configuration>
复制代码4, registered in FastPushAction in LogServer.java
public static void start() {
//注册action
ServerSetting.setAction(PushLogAction.class);
ServerSetting.setAction(FastPushAction.class);
try {
new LogServer().start(ServerSetting.getPort());
} catch (InterruptedException e) {
log.error("LoServer start error!", e);
}
}
复制代码performance
4-core single-instance 2G data size 2k, pressure measurement command is as follows:
ab -n2000000 -c500 "http://****:9001/log-server/fast_push?debugMode=0&sdk=tracker-view.js&ver=1.1.1&d_i=2020021955e4d920&url=https%3A%2F%2Fb.scene.eprezi.cn%2Fs%2FDPawp3qi%3Fshare_level%3D10%26from_user%3D20200211285b0f8f%26from_id%3Db9509c4e-7%26share_time%3D1581400182382%26from%3Dsinglemessage%26isappinstalled%3D0%26adpop%3D1&tit=%E6%B5%B7%E8%89%BA-%E5%8C%97%E4%BA%AC%E8%88%9E%E8%B9%88%E5%AD%A6%E9%99%A2%E4%B8%AD%E5%9B%BD%E8%88%9E%E8%80%83%E7%BA%A7%E6%95%99%E6%9D%90&ref=&u_a=&bro=%E5%BE%AE%E4%BF%A1&os=Android&o_v=8.1.0&eng=Webkit&man=Xiaomi&mod=HM-6&sns=weixin-singlemessage&n_t=wifi&s_i=v3x20200219eb3eadcb&c_i=da1c0dd6ad3b19b9a86f42bdfd31c69a&u_i=&c_p=Android&b_v=2.0&c_e=0.0.1&product=traffic_view&b_t=traffic&x_t=0&wx_o_i=&wx_n_n=&wx_sex=&wx_pro=&wx_cit=&wx_cou=&wx_hea=&wx_u_i=&wx_r_f=singlemessage&scene_id=121973512&scene_c_u=ff80808155520087015556e924cc0e8e&scene_code=DPawp3qi&scene_bizType=0&scene_property_eqAdType=1&scene_ext_yqc_ad=121973512&scene_member_type=&scene_user_type=1&foto_id=&foto_code=&rdt=1&domain=b.scene.eprezi.cn&media_id=1&pid=10000&works_id=121973512&earnings_user_id=ff80808155520087015556e924cc0e8e&publisher_id=24996&ad_unit_id=12&plan_id=201904091007&strategy_id=515&task_id=5156899&ad_platform=gdt&ad_type=tpl&ad_style_id=43&ad_size=&works_open_type=1&ad_page_num=0&man_2=%E5%B0%8F%E7%B1%B3&cou=&pro=&cit=&sex=0&new_user=&is_auth=0&device_size=36&d_t=1&e_t=element_view&count=1&scene_page_curr=&target_url=&conversion_type=ad_position_request&c_t=1582102328147"
复制代码Pressure test results:
Concurrency Level: 500
Time taken for tests: 83.936 seconds
Complete requests: 2000000
Failed requests: 0
Write errors: 0
Total transferred: 272000000 bytes
HTML transferred: 0 bytes
Requests per second: 23827.75 [#/sec] (mean)
Time per request: 20.984 [ms] (mean)
复制代码
Author: Yi, director of enterprise data show large -Yarn