Cogs and Levers A blog full of technical stuff

Getting started with Berkeley DB

In today’s post, I’m just going to gloss over some top level details in developing applications that use Berkeley DB.

What is it?

The first line of the Wikipedia article for Berkeley DB sums the whole story up pretty well, I think:

Berkeley DB (BDB) is a software library that provides a high-performance embedded database for key/value data.

Pretty simple. Berkeley DB is going to offer you an in-process database to manage data in your applications in a little more organised approach.

Getting installed

I’m using Debian Linux, more specifically the testing release. I’m sure the installation process will translate for other Debian releases and/or other Linux distributions with their respective package managers.

If you’ve already got a standard development/build environment running, you won’t need the build-essential package listed below.

sudo apt-get install build-essential libdb-dev

Building applications

Once you’ve finished writing an application using this library, you’ll need to link the Berkeley DB library against your application. This is all pretty simple as well:

gcc -ldb yourapp.o -o yourapp

The key piece being the -ldb linker library switch.

Some simple operations

The following blocks of code are heavily based off of the information contained in this pdf. That pdf has a heap of information in it well worth the read if you’re going to do something serious with Berkeley DB.

First thing you need to do, is to initialize the database structure that you’ll use to conduct all of your operations.

DB *db;
int ret;

/* setup the database memory structure */
ret = db_create(&db, NULL, 0);

db_create will allocate and fill out the structure of the DB typed pointer. It just sets it up ready for use. Once you’ve created the database handle, it’s time to actually open up a database (or create one).

ret = db->open(
	db, 
	NULL, 
	"test.db",
	NULL,
	DB_BTREE,
	DB_CREATE,
	0
);

open is going to try and find the requested database; (in my case test.db) and open it up. Failing that, it’ll create it (because of DB_CREATE). The format parameter is quite interesting. In the sample above, DB_BTREE has been specified. Looking at the DB->open() documentation:

The currently supported Berkeley DB file formats (or access methods) are Btree, Hash, Queue, and Recno. The Btree format is a representation of a sorted, balanced tree structure. The Hash format is an extensible, dynamic hashing scheme. The Queue format supports fast access to fixed-length records accessed sequentially or by logical record number. The Recno format supports fixed- or variable-length records, accessed sequentially or by logical record number, and optionally backed by a flat text file.

This gives you the flexability to use the format that suits your application best.

Once the database is open, writing data into it is as easy as specifying a key and value. There are some further data structures that need to be filled out, but the code is pretty easy to follow:

/* source data values */
char *name = "John Smith";
int id = 5;

DBT key, value;

memset(&key, 0, sizeof(DBT));
memset(&value, 0, sizeof(DBT));

/* setup the key */
key.data = &id;
key.size = sizeof(int);

/* setup the value */
value.data = name;
value.size = strlen(name) + 1;

/* write it into the database */
ret = db->put(
	db, 
	NULL, 
	&key, 
	&value, 
	DB_NOOVERWRITE
);

As the data member of the DBT struct is typed out as void *, you can store any information you’d like in there. Of course, you must specify the size so the system knows how much to write.

Reading values back out into native variables is just as easy:

/* destinations */
char read_name[256];
int read_id = 5;

memset(&key, 0, sizeof(DBT));
memset(&value, 0, sizeof(DBT));

/* setup the key to read */
key.data = &read_id;
key.size = sizeof(int);

/* setup the value to fill */
value.data = read_name;
value.ulen = 256;
value.flags = DB_DBT_USERMEM;

db->get(
	db, 
	NULL, 
	&key, 
	&value, 
	0
);

Finally, you’ll want to close your database once you’re done:

if (db != NULL) {
	db->close(db, 0);
	db = NULL;
}

Wireshark installation on Debian

A really quick guide on installing Wireshark on Debian.

The installation itself is pretty straight forward, however there is a little bit of reconfig work and user administration to get going for non-root users.

Install Wireshark from apt

$ sudo apt-get install wireshark

Reconfigure the wireshark-common package making sure to answer yes to the question asked.

$ sudo dpkg-reconfigure wireshark-common 

Add any user to the wireshark group that needs to be able to capture data off the network interfaces.

$ sudo usermod -a -G wireshark $USER

Remember, if you added yourself to this group; you’ll need to logout and log back in for the group changes to take effect.

Installing Eclipse Luna on Debian

A really quick guide on installing Eclipse Luna on Debian.

If you’re on a fresh machine, and you’re downloading/installing Eclipse for the purposes of Java development, you’ll want to install the JDK. To get this going, I install the openjdk-7-jdk out of the apt repository.

$ sudo apt-get install openjdk-7-jdk

After that finishes, or while, grab a copy of the Eclipse version that you need from the download page. Once it’s down, I normally extract it and then put it in a system-wide location (as opposed to just running it from my home directory).

$ tar -zxvf eclipse-*.tar.gz
$ sudo mv eclipse /opt

One little oddity before starting Eclipse up, I’ve had to apply a GTK setting. Prior to making this setting, Eclipse would crash!

Add the following lines to you /opt/eclipse/eclipse.ini file. Make sure it appears before the --launcher.appendVmargs directive.

--launcher.GTK_version
2

Flask deployment with nginx and uwsgi

Taking your applications from the development web server into a full application server environment is quite painless with nginx, uwsgi and virtualenv. This post will take you through the steps required to get an application deployed.

Server Setup

First of all, you’ll need to get your server in a state where it’s capable of serving HTTP content as well as housing your applications. If you’ve already got a server that will do this, you can skip this.

$ sudo apt-get install nginx uwsgi uwsgi-plugin-python python-dev python-setuptools build-essential
$ sudo easy_install pip
$ sudo pip install virtualenv

This will put all of the software required onto the server to house these applications.

Application setup with uWSGI

Each uWSGI application’s configuration is represented on the filesystem as an ini file, typically found in /etc/uwsgi/apps-available. Symlinks are established between files in this directory into /etc/uwsgi/apps-enabled to tell the uwsgi daemon that an application needs to be running.

The following is an example uWSGI configuration file that you can use as a template:

[uwsgi]
vhost = true
chmod-socket = 666
socket = /tmp/app.sock
plugins = python
venv = /path/to/proj/env
chdir = /path/to/proj
module = modulename
callable = app

This will get our application housed by uWSGI. You can now enable this application:

$ sudo ln -s /etc/uwsgi/apps-available/app.ini /etc/uwsgi/apps-enabled/app.ini
$ sudo service uwsgi restart

Web server setup

Finally, we’ll get nginx to provide web access to our application. You may have specific web site files that you need to modify to do this, but this example assumes that you’re in control of the default application.

Add the following section to /etc/nginx/sites-available/default:

location /app {
        include uwsgi_params;
        uwsgi_param SCRIPT_NAME /app;
        uwsgi_modifier1 30;
        uwsgi_pass unix:/tmp/app.sock;
}

Reload your web server config, and you’re ready to go:

$ sudo service nginx restart

PhoneGap Setup on Arch Linux

Here’s a few notes to getting PhoneGap up and running on an Arch Linix installation.

Dependencies

PhoneGap itself relies on some java tools, so you’ll need a jdk and ant.

$ sudo pacman -S jdk7-openjdk
$ sudo pacman -S apache-ant

In order to run your applications in an android simulator, you’ll need the android sdk installed. For the next steps, you’ll need to ensure that multilib is enabled in your /etc/pacman.conf file.

You’ll need the following packages installed from AUR:

After these have been successfully installed, using the suggested installation procedure guidance on the wiki, you’ll need to put these tools on your path:

$ export PATH=$PATH:/opt/android-sdk/tools:/opt/android-sdk/platform-tools:/opt/android-sdk/build-tools

PhoneGap is installed using npm which is part of the NodeJS suite, so you’ll need to have it installed as well:

$ sudo pacman -S nodejs

Installation

From the PhoneGap installation guide, installation should be just:

$ sudo npm install -g phonegap

Device Setup

Before you can run any applications, you’ll need to setup a device. No cpu images are installed by default, so you’ll need to install these first using the android command.

After installing the appropriate images, you can create a device using android avd.