Export Prometheus metrics from arbitrary unstructured log data.

About Grok

Grok is a tool to parse crappy unstructured log data into something structured and queryable. Grok is heavily used in Logstash to provide log data as input for ElasticSearch.

Grok ships with about 120 predefined patterns for syslog logs, apache and other webserver logs, mysql logs, etc. It is easy to extend Grok with custom patterns.

The grok_exporter aims at porting Grok from the ELK stack to Prometheus monitoring. The goal is to use Grok patterns for extracting Prometheus metrics from arbitrary log files.

How to run the example

Download grok_exporter-$ARCH.zip for your operating system from the releases page, extract the archive, cd grok_exporter-$ARCH, then run

./grok_exporter -config ./example/config.yml

The example log file exim-rejected-RCPT-examples.log contains log messages from the Exim mail server. The configuration in config.yml counts the total number of rejected recipients, partitioned by error message.

The exporter provides the metrics on http://localhost:9144/metrics:

Configuration

Example configuration:

global:
  config_version: 3
input:
  type: file
  path: ./example/example.log
  readall: true
imports:
- type: grok_patterns
  dir: ./logstash-patterns-core/patterns
metrics:
- type: counter
  name: grok_example_lines_total
  help: Counter metric example with labels.
  match: '%{DATE} %{TIME} %{USER:user} %{NUMBER}'
  labels:
    user: '{{.user}}'
    logfile: '{{base .logfile}}'
server:
  port: 9144

CONFIG.md describes the grok_exporter configuration file and shows how to define Grok patterns, Prometheus metrics, and labels. It also details how to configure file, stdin, and webhook inputs.

How to build from source

In order to compile grok_exporter from source, you need

• Go installed and $GOPATH set.
• gcc installed for cgo. On Ubuntu, use apt-get install build-essential.
• Header files for the Oniguruma regular expression library, see below.

Installing the Oniguruma library on OS X

brew install oniguruma

Installing the Oniguruma library on Ubuntu Linux

sudo apt-get install libonig-dev

Installing the Oniguruma library from source

curl -sLO https://github.com/kkos/oniguruma/releases/download/v6.9.5_rev1/onig-6.9.5-rev1.tar.gz
tar xfz onig-6.9.5-rev1.tar.gz
cd onig-6.9.5
./configure
make
make install

Installing grok_exporter

git clone https://github.com/fstab/grok_exporter
cd grok_exporter
git submodule update --init --recursive
go install .

The resulting grok_exporter binary will be dynamically linked to the Oniguruma library, i.e. it needs the Oniguruma library to run. The releases are statically linked with Oniguruma, i.e. the releases don't require Oniguruma as a run-time dependency. The releases are built with hack/release.sh.

Note: Go 1.13 for Mac OS has a bug affecting the file input. It is recommended to use Go 1.12 on Mac OS until the bug is fixed. Go 1.13.5 is affected. golang/go#35767.

More Documentation

User documentation is included in the GitHub repository:

• CONFIG.md: Specification of the config file.
• BUILTIN.md: Definition of metrics provided out-of-the-box.

Developer notes are available on the GitHub Wiki pages:

• tailer (tail -f)
• About the Regular Expression Library

External documentation:

• Extracting Prometheus Metrics from Application Logs - https://labs.consol.de/...
• Counting Errors with Prometheus - https://labs.consol.de/...
• [Video] Lightning talk on grok_exporter - https://www.youtube.com/...

Contact

• For feature requests, bugs reports, etc: Please open a GitHub issue.
• For bug fixes, contributions, etc: Create a pull request.
• Questions? Contact me at fabian@fstab.de.

Related Projects

Google's mtail goes in a similar direction. It uses its own pattern definition language, so it will not work out-of-the-box with existing Grok patterns. However, mtail's RE2 regular expressions are probably more CPU efficient than Grok's Oniguruma patterns. mtail reads logfiles using the fsnotify library, which might be an obstacle on operating systems other than Linux.

License

Licensed under the Apache License, Version 2.0. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0.