eris - Eris is the Greek Goddess of Chaos


version 0.007


eris exists to transform unstructured, chaotic log data into structured messages.

Born out of disappointment and regret of existing solutions like Logstash, fluentd, and their kind, eris aims to make development and debugging of parsers easy and transparent. The goal is to provide a config that be used to to index logging data into Elasticsearch while being flexible enough to work with log files on the system. This makes it friendly to approach from a maintenance perspective as we don't need to run a massive app to figure out how a log message will be restructured.


eris is structured to be flexible, extensible, and visible in every component.



Decoders are pluggable thanks to eris::role::pluggable and they are searched for in the the default namespace eris::log::decoder. To add other namespaces, use the search_path parameter in a config file:

        - 'my::app::decoder'

Decoders operate on the raw string and provide rudimentary key/value pairs for the other contexts to operate on. Unlike the contexts, every discovered decoder is run for every message.



Class providing access to installed and configured decoders on the system.


Class which uses the decoders to transform the raw data into structured data.


The abstract role which implements a decoder.

eris::log::decoder::syslog, eris::log::decoder::json

Default implementations of decoders.


Contexts are pluggable and are searched for in the default namespace eris::log::decoder. To add your own namespaces, use the search_path parameter in your config file:

        - 'my::app::context'

Contexts implement the interface documented in eris::role::context. There are 4 major things to consider when implementing a new context.


This method is called when the context matches the event data. This is where you can implement your own parsing or analysis of the event data. To add context to an event, use the eris::log's add_context() method. That context data will be available to future contexts.


Return an array of sample messages. This provides future developers with some data to use in testing and enhancing your context.


This specifies the field or fields that a matcher will operate on. There are two special fields * and _exists_. The * is used in conjunction with a matcher of * to match all messages. The _exists_ operator is used to check for the existence of a key in the context. A sample use of this field specifier is used by the eris::log::context::GeoIP context with an regex matcher to operate on any event data with field names matching '_ip$'.


Can be *, a string, a regex ref, an array reference, or a code reference. If matcher and field are set to *, every message matches. If a literal string, or array reference, the literal string is checked against the value of in the field specified above and returns 1 if they are equivalent. If a regex reference, the regex is applied to the value in the specified field and the context is applied if the regex matches. A code reference should return 1 if the event is relevant to the context and 0 if it doesn't apply.

The default field is 'program', and the default matcher is a string with the value equal to the context's name attribute. For instance, eris::log::context::sshd defaults it's name to 'sshd', and since it doesn't override the field, this context is only applied to events with a 'program' key with a value of 'sshd'.



Class providing access to installed and configured contexts on the system.


Class which uses the contexts to transform the raw data into structured data.


The abstract role which implements a context.

eris::log::context::sshd, eris::log::context::GeoIP

Selected example contexts


Dictionaries are used in conjunction with schemas to filter eris::log contexts down to only the keys and values we want. This allows better control of the data headed into storage to prevent key space explosions.



Class providing access to installed and configured dictionaries on the system.


Class which uses the dictionaries to filter structured data into a document.


The abstract role which implements a dictionary.

eris::dictionary::cee, eris::dictionary::eris::debug

Selected example contexts


Schemas perform the transformation from structured data into documents for indexing. They allow control of the structure and destination of the document being indexed.



Class providing access to installed and configured schemas on the system.


The abstract role which implements a schema.


Selected example contexts


The goal of eris is to provide a set of tools that can be glued together to transform unstructured logging data into structured data and then rules for taking that structured data and storing it somewhere. That sounds cool, but there's nothing useful about it unless you can start playing with it now.

This is why eris ships with sample implementations.


Here's a list of the scripts installed along with eris so you can start breaking things.

This script allows you to do a few useful things. To see what happens to unstructured data, you can try performing some simple transforms via the built-in sample_messages: --sample sshd

If you'd like to see what those samples look like as ElasticSearch build requests, you can: --sample sshd --bulk

Without the --sample argument, you can feed data to it using STDIN or a file as it'll use the Perl magic diamond to read data until an EOF is reached.

To see what the bulk output would look like from a few sources:

Via pipe:

    tail /var/log/messages | -b

Via a file: -b /var/log/messsages

Via STDIN for testing or manually importing data: -b

The script provides more options, pull up it's help with: --help

This script allows you to query the eris::dictionary to see what it knows about a particular field. src_ip

This is a sample implementation that performs indexing of data received over syslog to an ElasticSearch cluster. It will parse all messages passed to it over STDIN and send them to an ElasticSearch cluster. It's single threaded, so it won't be able to keep up with a full speed log load. See it's help output for options and details: --help

This is a wrapper around using POE::Component::WheelRun::Pool to provide a pool of workers for processing log data at scale. To use it with syslog-ng:

    destination d_eris { program("/usr/local/bin/" keep-alive(); ); };
    log  { source(src_network); destination(d_eris); };

See it's help output for options: --help

This is a wrapper around for use in environments using the POE::Component::Server::eris syslog service. This service is a simple, stateless syslog message dispatch system used primarily for development of new syslog parser use cases. It transforms the syslog stream into a subscription service any user on the local system can tap. IF you're using that server, you can run the to leverage POE::Component::Client::eris to receive messages from the upstream and dispatch them to a worker pool of's. For more information, see: --help


Brad Lhotsky <>


This software is Copyright (c) 2015 by Brad Lhotsky.

This is free software, licensed under:

  The (three-clause) BSD License