Skip to main content

Autonomous Log Collector and Observer

Project description

ALCO - Autonomous Log Collector and Observer

[![PyPI version](](

What's the problem

There is a widely used stack of technologies for parsing, collecting and
analysing logs - [ELK Stack](
It has very functional web interface, search cluster and a log transformation tool. Very cool, but:

* It's Java with well-known requirements for memory and CPUs
* It's ElasticSearch with it's requirements for disk space
* It's nodejs-based Logstash witch suddenly stops processing logs in some conditions.
* It's Kibana with very cool RICH interface which looses on all counts to `grep` and `less` in a task of log reading and searching.

Introducing ALCO

ALCO is a simple ELK analog which primary aim is to provide a online replacement for `grep` and `less`. Main features are:

* Django application for incident analysis in distributed systems
* schemeless full-text index with filtering and searching
* configurable log collection and rotation from RabbitMQ messaging server
* not a all-purpose monster

Technology stack

Let's trace log message path from some distributed system to ALCO web interface.

1. Python-based project calls `logger.debug()` method with text 'hello world'
2. At startup time [Logcollect]( library automatically configures python logging (or even [Django]( and [Celery]( one's) to send log messages to RabbitMQ server in JSON format readable both with ELK and ALCO projects.
3. ALCO log collector binds a queue to RabbitMQ exchange and processes messages in a batch.
4. It uses Redis to collect unique values for filterable fields and SphinxSearch to store messages in a realtime index.
5. When a message is inserted to sphinxsearch, it contains indexed `message` field, timestamp information and schemeless JSON field named `js` with all log record attributes sent by python log.
6. Django-based web interface provides API and client-side app for searching collected logs online.


* Python 2.7 or 3.3+
* [Logcollect]( for python projects which logs are collected
* [RabbitMQ]( server for distributed log collection
* [SphinxSearch]( server 2.3 or later for log storage
* [Redis]( for SphinxSearch docid management and field values storage
* [django-sphinxsearch]( as a database backend for `Django>=1.8` (will be available from PyPI)


1. You need to configure logcollect in analyzed projects (see [README]( If RabbitMQ admin interface shows non-zero message flow in `logstash` exchange - "It works" :-)

2. Install alco and it's requirements from PyPi

pip install alco

3. Next, create django project, add `sphinxsearch` database connection and configure `` to enable alco applications

# For SphinxRouter

'ENGINE': 'sphinxsearch.backend.sphinx',
'HOST': '',
'PORT': 9306,

# Auto routing log models to SphinxSearch database

'rest_framework', # for API to work

ROOT_URLCONF = 'alco.urls'

4. Configure ALCO resources in ``:

# log messaging server
'host': '',
'userid': 'guest',
'password': 'guest',
'virtual_host': '/'

# redis server
'REDIS': {
'host': '',
'db': 0
# url for fetching sphinx.conf dynamically
# name of django.db.connection for SphinxSearch
# number of results in log view API

5. Run `syncdb` or better `migrate` management command to create database tables.

6. Run webserver and create a LoggerIndex from [django admin](

7. Created directories for sphinxsearch:


8. Next, configure sphinxsearch to use generated config:


searchd -c

`` is a simple script that imports `alco.sphinx_conf` module which fetches generated `sphinx.conf` from alco http api and created directories for SphinxSearch indices:


# coding: utf-8
import os
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'settings')

from alco import sphinx_conf

9. Run log collectors:

python start_collectors --no-daemon

If it shows number of collected messages periodically - then log collecting is set up correctly.

10. Configure system services to start subsystems automatically:

* nginx or apache http server
* django uwsgi backend
* alco collectors (`start_collectors` management command)
* sphinxsearch, redis, default database for Django

11. Open `<logger_name>/` to read and search logs online.


We successfully configured SphinxSearch to use python from `virtualenv`, adding some environment variables to start script (i.e. FreeBSD rc.d script):


sphinxsearch_prestart ()
# nobody user has no HOME
export PYTHON_EGG_CACHE=/tmp/.python-eggs
# python path for virtualenv interpreter should be redeclared
export PYTHONPATH=${venv_path}/lib/python3.4/:${venv_path}/lib/python3.4/site-packages/
. "${virtualenv_path}/bin/activate" || err 1 "Virtualenv is not found"
echo "Virtualenv ${virtualenv_path} activated: `which python`"



In this case _shebang_ for `` must point virtualenv's python interpreter.

Production usage

For now ALCO stack is tested in preproduction environment in our company and is actively developed. There are no reasons to say that it's not ready for production usage.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
alco-0.6.5.tar.gz (99.9 kB) Copy SHA256 hash SHA256 Source None

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page