Sherlock is an open source infrastructure monitoring application.
As of now, its in the super-very-early stages of development, but some feature highlights will be:
- Server/Agent architecture where agents installed on the servers will push metrics to the Sherlock server.
- Gathering of all types of data from agents: process information, load information, various data gathered from /proc, network status, contents of monitored log files and data gathered by custom plugins.
- Alerting built on user definable criterias against the data provided to Sherlock by the agents.
- Indexed and searchable log data from each node.
It will also be using file system based configuration files to set up everything instead of a database-oriented and UI-administerable configuration system, to make Sherlock easy to configure and administer using configuration management tools like Chef. If you have an infrastructure you care enough about to monitor it, you should care enough to use Chef or any other similar tools as well. A Chef cookbook will be built for Sherlock after the initial implementation phase has completed and we have a working beta release.
The Sherlock server itself is be written in Ruby utilizing EventMachine and Sinatra. Metrics are stored in MongoDB.
System agents are implemented in Ruby in a separate project.