/lapidary

Search engine for logs stored in PostgreSQL

Primary LanguageClojureGNU General Public License v3.0GPL-3.0

Lapidary

A search engine for structured data using Postgres, with a similar syntax to Apache Lucene. This is intended for small to medium sized centralized logging configurations.

Pull requests welcome, especially documentation!

CircleCI

Features

  • Data is stored as JSON in PostgreSQL
  • Lucene like search syntax
  • Full text search

Screenshots

Tables

Tables

Query

Query

Installation

Docker

Using the docker-compose.yml file from the repository, run:

docker-compose up

This will start PostgreSQL, Lapidary and fluentd. You can log into Lapidary at http://localhost:8081 using the username admin and the password "ChangeMe!". Fluentd is listening on 24230 for forward connections, and 5145 for syslog.

You can log a message using:

logger -n localhost -P 5145 "Hello Lapidary!"

Configuration

Reference

Query String

Querying the database is loosely based on Lucene syntax, though grouping is done in a more lisp like style.

The query consists fields to search, and the values to search for. If your value contains whitespace, you can wrap it in quotation marks. For example, to find all records which contain the word fail in the message field, you could use:

message:fail

By default, each field you search is joined together logically using or, so the following will find records which have fail or ssh in the `message field:

message:fail message:ssh

This could also be written using a list of acceptable values:

message:(fail ssh)

You can build more complex searches using and and or, with or being the default, so that the above search will logically expand to be:

(or message:fail message:ssh)

If you were looking for records that match both fail and ssh:

(and message:fail message:ssh)

A term can be negated by prepending it with !:, so to find records with fail and ssh, but not mysql in message, you could use:

(and message:fail message:ssh !message:mysql)

You can specify a comparison operator, using one of <, <=, >, >=, =, !=. For example, to find records with OriginStatus between 400 and 499, you could use:

(and OriginStatus:>=400 OriginStatus:<=499)

For = and != the value will be matched exactly. For example, to match records with a RequestMethod of POST but not POSTAL, you can use:

RequestMethod:=POST

Addtionally, there are regex operators of ~ for a case sensitve regex, ~* for a case insensntive regex, and the negative form of both using !~ or !~*. If you want to match the regex /sockjs/.*/websocket$ in a RequestPath field, you could use:

RequestPath:~"sockjs/.*/websocket$"

The comparison operators can also be used in a list:

RequestMethod:(=POST =HEAD)

As an alternative to using and, you can use a range of values with [] used to signify both ends of the range being inclusive, {} for exclusive, and [} or {] as a mix of both. The values are seperated using either .. or to. For OriginStatus between 400 and 499, you could use:

OriginStatus:[400 to 499]

Or, for the same effect:

OriginStatus:[400..500}

API

TODO: Currently not available

Project Roadmap