Although this framework is fully functional and stable, since it was released Splunk have now created their own Java Logging Library. So I recommend that you use the formally Splunk developed and supported offering that can be found here : http://dev.splunk.com/view/splunk-logging-java/SP-CAAAE2K
The purpose of this project is to create a logging framework to allow developers to as seamlessly as possible integrate Splunk best practice logging semantics into their code and easily send their log events to Splunk. There are also custom handler/appender implementations and config examples for the most prevalent Java logging frameworks in play.
- LogBack
- Log4j 1.x
- Log4j 2
- java.util logging
This framework contains :
- java.util.logging handler for logging to Splunk REST endpoints
- java.util.logging handler for logging to Splunk Raw TCP Server Socket
- java.util.logging handler for logging to Splunk HEC Endpoint
- Log4j appender for logging to Splunk REST endpoints
- Log4j appender for logging to Splunk Raw TCP Server Socket
- Log4j appender for logging to Splunk HEC Endpoint
- Logback appender for logging to Splunk REST endpoints
- Logback appender for logging to Splunk Raw TCP Server Socket
- Logback appender for logging to Splunk HEC Endpoint
- Log4j 2 appender for logging to Splunk HEC Endpoint
- Log4j 2 config examples for TCP and UDP logging
- Example logging configuration files for all the above
- Javadocs
If you want to use UDP to send events to Splunk , then Log4j 1.x and Logback already have Syslog Appenders. Log4j 2 has a UDP Appender and Syslog Appender. And of course you can still use any File appenders and have the file monitored by a Splunk Universal Forwarder.
I generally recommend using the raw TCP or HEC handlers/appenders I have provided , they perform the best, and have features coded into them for auto connection re-establishment and configurable buffering of log events which will get flushed upon reconnection.
Log4j 2 and Log4j 1.x are very distinct from one another. Logback was actually the "new version" of Log4j 1.x , and then Log4J 2 attempted to improve upon Logback. This rather convoluted family tree has essentially transpired with 3 different logging frameworks in play, each with different characteristics. Log4j 1.x still has a very large legacy usage base in enterprise software therefore warrants addressing with its own custom appenders and example configurations.
I always advocate the best practice of using a Splunk Universal Forwarder(UF) monitoring local files wherever possible. Not only do you get the features inherent in the UF, but you get the added resiliency of the persistence of files. However, there are going to be situations where, for whatever reason(technical or bureaucratic), that a UF can not be deployed.In this case, Splunk Java Logging can be used to forward events to Splunk. Furthermore, in either scenario, you can still utilize the SplunkLogEvent class to construct your log events in best practice semantic format.
The HTTP REST ,Raw TCP and HEC handler/appenders have autonomous socket reconnection logic in case of connection failures. There is also internal event queuing that is loosely modelled off Splunk's outputs.conf for Universal Forwarders. You can set these propertys :
- maxQueueSize : defaults to 500KB , format [integer|integer[KB|MB|GB]]
- dropEventsOnQueueFull : defaults to false , format [ true | false]
And you can use a parallel File appender if you absolutely need disk persistence.
If you want "data cloning" functionality, then you can leverage the logging configuration and have (n) different appender definitions for your various target Indexers.
If you wish to have load balancing of your log events, then configure your logging appenders to send to a Splunk Universal Forwarder acting as a load balancing intermediary before you Indexer Cluster.
Log4J 2 has a Failover appender you can use : http://logging.apache.org/log4j/2.x/manual/appenders.html#FailoverAppender There is an example in config/log4j2.xml
Log4J 2 has a Routing appender you can use : http://logging.apache.org/log4j/2.x/manual/appenders.html#RoutingAppender
Log4j and Logback are thread safe.
The Splunk Java Logging Framework is licensed under the Creative Commons 3.0 License. Details can be found in the file LICENSE.
- Untar releases/splunklogging-1.3.tar.gz
- All the required jar files are in the lib directory..
- Assume you know how to setup your classpath to use your preferred logging framework implementation.
- There is a simple code example here https://github.com/damiendallimore/SplunkJavaLogging/blob/master/src/com/splunk/logging/examples/Example.java
- There are sample logging config files in the config directory for the 4 logging frameworks
If you haven't already installed Splunk, download it here: http://www.splunk.com/download. For more about installing and running Splunk and system requirements, see Installing & Running Splunk (http://dev.splunk.com/view/SP-CAAADRV).
Get the Splunk Java Logging Framework from GitHub (https://github.com/) and clone the resources to your computer. For example, use the following command:
git clone https://github.com/damiendallimore/SplunkJavaLogging.git
This project was initiated by Damien Dallimore
| ddallimore@splunk.com | |
| @damiendallimore | |
| Splunkbase.com | damiend |