同事整理的，在此分享。logback，log4j2 等slf4j的日志实现都可以以json格式输出日志， 这里采用的是logback。当然也可以以文本行的格式输出，然后在logstash里通过grok解析，但是直接以json格式输出，在logstash处理时效率会高一点。
Logback 输出 Json格式日志文件
为了让 logback 输出JSON 格式的日志文件，需要在pom.xml 加入如下依赖
<dependency> <groupId>net.logstash.logback</groupId> <artifactId>logstash-logback-encoder</artifactId> <version>4.8</version> <scope>runtime</scope></dependency> |
logback日志配置示例
<appender name="errorFile" class="ch.qos.logback.core.rolling.RollingFileAppender"> <filter class="ch.qos.logback.classic.filter.LevelFilter"> <level>ERROR</level> <onMatch>ACCEPT</onMatch> <onMismatch>DENY</onMismatch> </filter> <file>${log.dir}/elk/error.log</file> <!-- 当前的日志文件文件放在 elk文件下，该日志的内容会被filebeat传送到es --> <rollingPolicy class="ch.qos.logback.core.rolling.TimeBasedRollingPolicy"> <! -- 历史日志会放到 bak 文件下，最多保存7天的历史，最多占用 1G的空间 --> <fileNamePattern>${log.dir}/bak/error.%d{yyyy-MM-dd}.log</fileNamePattern> <maxHistory>7</maxHistory> <totalSizeCap>1GB</totalSizeCap> </rollingPolicy> <encoder class="net.logstash.logback.encoder.LoggingEventCompositeJsonEncoder"> <providers> <pattern> <pattern> { "tags": ["errorlog"], "project": "myproject", "timestamp": "%date{"yyyy-MM-dd'T'HH:mm:ss,SSSZ"}", "log_level": "%level", "thread": "%thread", "class_name": "%class", "line_number": "%line", "message": "%message", "stack_trace": "%exception{5}", "req_id": "%X{reqId}", "elapsed_time": "#asLong{%X{elapsedTime}}" } </pattern> </pattern> </providers> </encoder></appender> |
Json 字段说明:
|
名称
|
说明
|
备注
|
|
|
|
|
|
|---|---|---|---|---|---|---|---|
| tags | 用于说明这条日志是属于哪一类日志 | ||||||
timestamp |
日志记录时间 | ||||||
project |
系统名称，该日志来自于哪个系统 | ||||||
log_level |
输出日志级别 | ||||||
thread |
输出产生日志的线程名。 | ||||||
class_name |
输出执行记录请求的调用者的全限定名 |
|
|||||
line_number |
输出执行日志请求的行号 |
|
|||||
message |
输出应用程序提供的信息 | ||||||
stack_trace |
异常栈信息 | ||||||
req_id |
请求ID，用于追踪请求 | 需要引入aop-logging | |||||
elapsed_time |
该方法执行时间,单位: 毫秒 | 需要引入aop-logging |
%X{key}: 表示该项来自于SLF4j MDC,需要引入 aop-logging
<dependency> <groupId>com.cloud</groupId> <artifactId>xspring-aop-logging</artifactId> <version>0.7.1</version></dependency>针对web应用，在 web.xml 中加入 ReqIdFilter,该过滤器会在MDC 加入 reqId<filter> <filter-name>aopLogReqIdFilter</filter-name> <filter-class>com.github.nickvl.xspring.core.log.aop.ReqIdFilter</filter-class></filter><filter-mapping> <filter-name>aopLogReqIdFilter</filter-name> <url-pattern>/*</url-pattern></filter-mapping>or register in springboot like this:@Beanpublic FilterRegistrationBean getDemoFilter(){ ReqIdFilter reqIdFilter=new ReqIdFilter(); FilterRegistrationBean registrationBean=new FilterRegistrationBean(); registrationBean.setFilter(reqIdFilter); List<String> urlPatterns=new ArrayList<String>(); urlPatterns.add("/*"); registrationBean.setUrlPatterns(urlPatterns); registrationBean.setOrder(100); return registrationBean;}如果需要记录该方法执行时间: elapsed_time，如果在该类或者方法上加入如下注解：import com.github.nickvl.xspring.core.log.aop.annotation.LogDebug;import com.github.nickvl.xspring.core.log.aop.annotation.LogInfo;@LogInfo // 当logger 设为level=INFO 会输出@LogException(value = {@Exc(value = Exception.class, stacktrace = false)}, warn = {@Exc({IllegalArgumentException.class})}) //当logger 设为level=error 会输出针对dubbo 消费者的日志记录,dubbo消费者是通过 javassist 生成的动态类型,如果要监控该dubbo接口的传入参数，返回值，和调用时间 需要引入aop-logging,以及在 eye-rpc包中的接口上给对应的类或方法 加上上面的注解。dubbo 消费者的日志会输出如下配置: <logger name="com.alibaba.dubbo.common.bytecode" level="INFO" additivity="false"> <appender-ref ref="dubboApiFile"/></logger> |
ElasticSearch 模板设置
curl -XPUT http://localhost:9200/_template/log -d '{ "mappings": { "_default_": { "_all": { "enabled": false }, "_meta": { "version": "5.1.1" }, "dynamic_templates": [ { "strings_as_keyword": { "mapping": { "ignore_above": 1024, "type": "keyword" }, "match_mapping_type": "string" } } ], "properties": { "@timestamp": { "type": "date" }, "beat": { "properties": { "hostname": { "ignore_above": 1024, "type": "keyword" }, "name": { "ignore_above": 1024, "type": "keyword" }, "version": { "ignore_above": 1024, "type": "keyword" } } }, "input_type": { "ignore_above": 1024, "type": "keyword" }, "message": { "norms": false, "type": "text" }, "offset": { "type": "long" }, "source": { "ignore_above": 1024, "type": "keyword" }, "tags": { "ignore_above": 1024, "type": "keyword" }, "type": { "ignore_above": 1024, "type": "keyword" } } } }, "order": 0, "settings": { "index.refresh_interval": "5s" }, "template": "log-*"}'curl -XPUT http://localhost:9200/_template/log-java -d '{ "mappings": { "_default_": { "properties": { "log_level": { "ignore_above": 1024, "type": "keyword" }, "project": { "ignore_above": 1024, "type": "keyword" }, "thread": { "ignore_above": 1024, "type": "keyword" }, "req_id": { "ignore_above": 1024, "type": "keyword" }, "class_name": { "ignore_above": 1024, "type": "keyword" }, "line_number": { "type": "long" }, "exception_class":{ "ignore_above": 1024, "type": "keyword" }, "elapsed_time": { "type": "long" }, "stack_trace": { "type": "keyword" } } } }, "order": 1, "settings": { "index.refresh_interval": "5s" }, "template": "log-java-*"}' |
logstatsh 设置
logstash-java-log
if [fields][logType] == "java" { json { source => "message" remove_field => ["offset"] } date { match => ["timestamp","yyyy-MM-dd'T'HH:mm:ss,SSSZ"] remove_field => ["timestamp"] } if [stack_trace] { mutate { add_field => { "exception_class" => "%{stack_trace}" } } } if [exception_class] { mutate { gsub => [ "exception_class", "
", "", "exception_class", ":.*", "" ] } }} |
filebeat 设置
filebeat.yml
filebeat.prospectors:- input_type: log paths: - /eyebiz/logs/eyebiz-service/elk/*.log # eyebiz-service 日志 - /eyebiz/logs/eyebiz-web/elk/*.log # eyebiz-web 日志 fields: logType: "java" docType: "log-java-dev" |