about // doc

MongoDB Quickstart

Created: Oct, 2015

Last modified: Sep, 2016

General Concepts

A list of MongoDB glossaries can be found here. But at its minimum, there are three terms: database, collection (like the “table” in a RBDMS) and document (the unit of a database, like the row in a RBDMS). For example:

  • A MongoDB server can have multiple databases, one database can have multiple collections, and a collection can have multiple documents. The the shell commands below to have a grasp of what it looks like.

  • MongodDB is schema-less (a feature of NoSQL), meaning that the documents in a collection don’t have to have the same structure.

Another feature of MongoDB is that a database or collection can be created on the fly, but its existence can only come into being when the first document under it is created.

Setting It Up

  • Install MongoDB on Ubuntu: doc from mongodb.org

  • MongoDB failed to start on Ubuntu, “errno:98 Address already in use for socket: “: restart mongod on Ubuntu to release the occupied port:

sudo service mongod restart
  • MongoDB needs space of > 3GB by default, or it will fail to start. In that case (for small VMs), start it with “–smallfiles” option:
mongod --smallfiles --dbpath /path/to/database

Quick-start MongoDB Shell Commands

  • To start MongoDB console/shell after the daemon mongod is started, do:
  • In the MongoDB shell, to list all databases, do:
show dbs
  • To create a new database or switch to another database named “datadump”, do
use datadump

Note that if it is a new database “datadump”, it will not be created until the first record is inserted below.

At this point, the shell variable “db” will be referenced to “datadump” (also see below).

  • To list all collections of the current database, do
show collections
  • To insert a document to the current database, do:
db.mycollection1.insert({"title": "user1", "email": "user1@example.com"})

this will create a collection named “mycollection1” if it hasn’t be created, and add a document to this collection. From this point, this collection can be referenced as “db.mycollection1” in the shell.

  • One can also create an array of records and insert like
newrec = [{"title": "user2", "email": "user2@example.com"}, {"title": "user3", "email": "user3@example.com"}]
  • To list all documents in a collection, do
  • To print the content of all documents in a clean JSON way
  • To remove all documents in a collection, do

A query is passed to the remove() function to match the target queries. See query operator for details.

Note that even all documents in a collection are removed, the collection still exists, and show up in the results of show collections.

  • To drop a collection (even if it is non-empty), do

This will decommission a collection from a database, making it disappear in the results of show collections.

  • To drop the current database, do
  • Press Ctrl+”d” to quit the console.

  • A list of commonly used MongoDB shell commands is here.

Using It with Python

MongoDB has a Python driver called PyMongo. First initiate a Virtualenv container:

virtualenv pymongo_env
. ./pymongo_env/bin/activate
pip install --upgrade pip

And install dependencies and pymongo

sudo apt-get install python-dev libffi-dev libssl-dev 
pip install requests[security]
pip install pymongo

Otherwise, there will be “InsecurePlatformWarning” after installing pymongo (see this post).

Refer to PyMongo documentation and tutorials on the usage.

Note that PyMongo is a blocking MongoDB driver, which may be fine for basic use. If non-blocking CRUD are needed, check out Motor.

Secure Remote Access of MongoDB

MongoDB listens only to localhost by default. One basic scenario is that I need to access a database running on Server A and listening 27017 from another Server B. I find the SSH tunneling method in this post a simple solution that is also secure enough. It can be further supplemented by autossh to keep the SSH tunnel alive. The following assumes the MongoDB on Server A uses default settings.

First install autossh

sudo apt-get install autossh

Now generate SSH public-private key pairs on Server B


and concatenate the public key on Server B to “~/.ssh/authorized_keys” on Server A. I leave the key passphrase empty for convenience.

Create an autossh tunnel on Server B:

autossh -M 0 -fN -L 4321:localhost:27017 UserA@ServerA

The command above will create a tunnel between the port 4321 of localhost on Server B to the port 27017 of localhost on Server A. Autossh has many powerful options that need further reading. If the public key of Server B is not at “~/.ssh/id_rsa.pub”, specify the correct path with -i path_to_pub_key_server_b option.

Now, one can access the database on Server A from Server B with port 4321 of localhost.

comments powered by Disqus