LogDNA Appender for Logback

LogDNA is a hosted logging platform: https://logdna.com

This small library provides an appender for logback (a popular logging subsystem for the JVM).

The appender pushes log entries to LogDNA via HTTPS

Logback's thread-bound storage, the MDC, is sent to LogDNA as metadata to be indexed and then searchable. (See screenshot, and more below)

How To

Logback uses XML file in known locations with the most common being classpath:/logback.xml

This is what a logback.xml might look like:

<?xml version="1.0" encoding="UTF-8"?>
<configuration>

    <!-- Take both of these LOGDNA appenders -->
    <!-- this one is a plain old HTTP transport -->
    <appender name="LOGDNA-HTTP" class="net._95point2.utils.LogDNAAppender">
    <appName>LogDNA-Logback-Test</appName>
    <ingestKey>${LOGDNA_INGEST_KEY}</ingestKey>
    <includeStacktrace>true</includeStacktrace>
    </appender>

    <!-- ... and this one should be attached to the root -->
    <appender name="LOGDNA" class="ch.qos.logback.classic.AsyncAppender">
      <appender-ref ref="LOGDNA-HTTP" />
    </appender>


  <appender name="CONSOLE" class="ch.qos.logback.core.ConsoleAppender">
    <encoder>
      <pattern>%m%n</pattern>
      <charset>utf8</charset>
    </encoder>
  </appender> 

    <root level="INFO">
        <appender-ref ref="CONSOLE" />
      <appender-ref ref="LOGDNA" /><!-- the async one -->
    </root>
</configuration>

This setup uses an asynchronous wrapper to deal with log shipping on another thread and has some sensible options for dealing with buffer management. Read more here

Get it using Maven

1. Add a link to this project's repository in your pom

project > repositories >

<repository>
    <snapshots>
        <enabled>false</enabled>
    </snapshots>
    <id>bintray-robshep-oss</id>
    <name>bintray</name>
    <url>http://dl.bintray.com/robshep/oss</url>
</repository>

2. Add the dependency

project > dependencies >

<dependency>
  <groupId>net._95point2.utils</groupId>
  <artifactId>logback-logdna</artifactId>
  <version>1.1.0</version>
</dependency>

Or, plain ol' download

Or just download the Jar and it's dependencies from https://github.com/robshep/logback-logdna/releases

Configure

You can go to town on most other logback configurations but the LogDNA only has a couple of settings (shown as the default)

• <appName>LogDNA-Logback-Test</appName> set this for good log management in LogDNA
• <ingestKey>${LOGDNA_INGEST_KEY}</ingestKey> signup to LogDNA and find this in your account profile
• <includeStacktrace>true</includeStacktrace> this library can send multiline stacktraces (see image) - Raw syslog transport cannot
• <sendMDC>true</sendMDC> copies over logback's Mapped Diagnostic Context (MDC) as LogDNA Metadata which are then indexed and searchable.
• <tags>production</tags> Sets the tags to send to LogDNA, multiple tags should be comma-separated

More Info

• The log line displays the thread, the logger (class) and the message
• LogDNA's metadata is populated with the logger as an indexable/searchable property.
• The HTTP Transport is done by the very lightweight DavidWebb REST Library and so doesn't introduce bulky dependencies

Using the MDC for Meta Data

The combination of logback's MDC and LogDNA's metadata support is pretty powerful and means you can correlate web-requests, or a userId, or something else that happens in a thread in your application.

For example, doing this anywhere in your application...

MDC.put("customerId", "C-1");
MDC.put("requestId", "cafebabe1");

... means that you can then go and search for say, that customer in logDNA like this:

meta.customerId:"C-1"

Screenshot of the LogDNA log viewer, including extra metadata