MongoLog: Centralized logging made simple using MongoDB

Info:	MongoDB python logging handler. Python centralized logging made easy.
Author:	Andrei Savu
Maintainer:	Jorge Puente Sarrín

Setup

Before using this handler for logging you will need to create a capped collection on the MongoDB server.

You can do this using the following commands in the MongoDB shell:

> use mongolog
> db.createCollection('log', {capped:true, size:100000})

... and you are ready. Running stats() function on log collection should show something like this:

> db.log.stats()
{
        "ns" : "mongolog.log",
        "count" : 0,
        "size" : 0,
        "storageSize" : 102400,
        "numExtents" : 1,
        "nindexes" : 1,
        "lastExtentSize" : 102400,
        "paddingFactor" : 1,
        "systemFlags" : 1,
        "userFlags" : 0,
        "totalIndexSize" : 8176,
        "indexSizes" : {
                "_id_" : 8176
        },
        "capped" : true,
        "max" : 2147483647,
        "ok" : 1
}

Usage

>>> import logging
>>> from mongolog.handlers import MongoHandler
>>>
>>> log = logging.getLogger('demo')
>>> log.setLevel(logging.DEBUG)
>>> log.addHandler(MongoHandler.to(db='mongolog', collection='log'))
>>>
>>> log.debug('Some message')

Check the samples folder for more details.

Why centralized logging?

Easy troubleshouting:
- Having the answers to why? quickly and accurately.
- For troubleshouting while the system is down.
- Removed risk of loss of log information.
Resource tracking.
Security.

What is MongoDB?

"MongoDB is a high-performance, open source, schema-free document-oriented database."

It can eficiently store arbitrary JSON objects. You can read more at MongoDB website.

Why MongoDB is great for logging?

MongoDB inserts can be done asynchronously.
Old log data automatically LRU's out thanks to capped collections.
It's fast enough for the problem.
Document-oriented / JSON is a great format for log information.

Read more about this subject on the MongoDB blog.