Is there a Logback Layout that Creates JSON Object

2020-05-21 09:52发布

问题:

I want to send log events to Loggly as JSON objects with parameterized string messages. Our project currently has a lot of code that looks like this:

String someParameter = "1234";
logger.log("This is a log message with a parameter {}", someParameter);

We're currently using Logback as our SLF4J backend, and Logback's JsonLayout to serialize our ILogEvent objects into JSON. Consequentially, by they time our log events are shipped to Loggly, they look like this:

{
    "message": "This is a log message with a parameter 1234",
    "level": INFO,
    ....
}

While this does work, it sends a different message string for every value of someParameter, which renders Loggly's automatic filters next to useless.

Instead, I'd like to have a Layout that creates JSON that looks like this:

{
    "message": "This is a log message with a parameter {}",
    "level": INFO,
    "parameters": [
        "1234"
    ]
}

This format would allow Loggly to group all log events with the message This is a log message with a parameter together, regardless of the value of someParameter.

It looks like Logstash's KV filter does something like this - is there any way to accomplish this task with Logback, short of writing my own layout that performs custom serialization of the ILogEvent object?

回答1:

You could use a Mapped Diagnostic Context to set a stamp for each of those type of log messages that you could then filter on once in loggly.

According to the source of JsonLayout the stamp is stored as a separate value in the JSON.



回答2:

There is a JSON logstash encoder for Logback, logstash-logback-encoder



回答3:

So for me I was trying to log execution times, I created a pojo called ExecutionTime with name, method, class, duration.

I was then able to create it:

ExecutionTime time = new ExecutionTime("Controller Hit", methodName, className, sw.getTotalTimeMillis());

For logging I then used:

private final Logger logger = LoggerFactory.getLogger(this.getClass());
logger.info(append("metric", time), time.toString());

Make sure you have:

import static net.logstash.logback.marker.Markers.append;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

This will log something like this:

{  
   "ts":"2017-02-16T07:41:36.680-08:00",
   "msg":"ExecutionTime [name=Controller Hit, method=setupSession, className=class com.xxx.services.controllers.SessionController, duration=3225]",
   "logger":"com.xxx.services.metrics.ExecutionTimeLogger",
   "level":"INFO",
   "metric":{  
      "name":"Controller Hit",
      "method":"setupSession",
      "className":"class com.xxx.services.controllers.SessionController",
      "duration":3225
   }
}

Might be a different set up as I was using logback-spring.xml to output my logs to json:

<?xml version="1.0" encoding="UTF-8"?>
<configuration>
    <include resource="org/springframework/boot/logging/logback/base.xml"/>
    <property name="PROJECT_ID" value="my_service"/>
    <appender name="FILE" class="ch.qos.logback.core.rolling.RollingFileAppender">
        <File>app/logs/${PROJECT_ID}.json.log</File>
        <encoder class="net.logstash.logback.encoder.LogstashEncoder">
            <fieldNames>
                <timestamp>ts</timestamp>
                <message>msg</message>
                <thread>[ignore]</thread>
                <levelValue>[ignore]</levelValue>
                <logger>logger</logger>
                <version>[ignore]</version>
            </fieldNames>
        </encoder>
        <rollingPolicy class="ch.qos.logback.core.rolling.FixedWindowRollingPolicy">
            <maxIndex>10</maxIndex>
            <FileNamePattern>app/logs/${PROJECT_ID}.json.log.%i</FileNamePattern>
        </rollingPolicy>
        <triggeringPolicy class="ch.qos.logback.core.rolling.SizeBasedTriggeringPolicy">
            <MaxFileSize>20MB</MaxFileSize>
        </triggeringPolicy>
    </appender>
    <logger name="com.xxx" additivity="false" level="DEBUG">
        <appender-ref ref="FILE"/>
        <appender-ref ref="CONSOLE"/>
    </logger>
    <root level="WARN">
        <appender-ref ref="FILE"/>
    </root>
</configuration>


回答4:

Here's a recently created project that provides a JSON-specific logging API and works with SLF4J:

https://github.com/savoirtech/slf4j-json-logger



回答5:

Like already answered you'll get a one-dimensional JSON tree with MDC and/or using a Marker with logstash-logback-encoder.

If you are also looking for the following:

  • codebooks for definition of logged datatype key and type,
  • configuration of log-aggregation tools (like elasticsearch)
  • generated Java helper-code for efficient and correct logging

then try a project I've created: json-log-domain. It defines a simple YAML-format definition from which the above can be generated.

An example helper-code statement would be

logger.info(host("localhost").port(8080), "Hello world");

while generated markdown would like something like this.