How to regenerate SSH client keys

Recently, a SSH client bug was discovered that could let the server read client private keys in some situations.

If you’re affected, then follow these steps. SSH regulars will be familiar with most of the commands used.

If you use keys

If you do use keys, you should regenerate them. Start by backing up your old key. Assuming it’s at the default location, just use-

$ mv ~/.ssh/id_rsa ~/.ssh/id_rsa.old
$ mv ~/.ssh/ ~/.ssh/

Next, create or edit ~/.ssh/config, and add the following line. This disables the roaming feature, which was part of the reason:

UseRoaming no

Generate new keys-

$ ssh-keygen 
Generating public/private rsa key pair.
Enter file in which to save the key (/home/example/.ssh/id_rsa): 
Enter passphrase (empty for no passphrase): 
Enter same passphrase again: 
Your identification has been saved in /home/example/.ssh/id_rsa.
Your public key has been saved in /home/example/.ssh/

Now copy the new key over, using the old keys. You would normally use ssh-copy-id, but it adds keys rather than replacing them, which is why I’ll do it manually-

scp -i ~/.ssh/id_rsa.old

SSH in with the old key:

ssh -i ~/.ssh/id_rsa.old

Replace all current authorized keys with the newly generated one:

mv -f ~/.ssh/authorized_keys
chmod 600 ~/.ssh/authorized_keys

Logout, log back in with the new key to make sure it works, and you’re done.

If you don’t use keys

If you don’t use keys for login, now could be a good time to start. Public-key authentication provides much stronger protection from brute force attacks, and also prevents the need for the server to be sent your password each time you log in.

Create or edit ~/.ssh/config as above, and make sure it contains this line-

UseRoaming no

Generate a keypair and add your public key to the server-


Test it out-


At his point, you should disable password-based authentication in OpenSSH server. If you don’t control the server, then just get a good random password for your account to prevent it being brute-orced:

openssl rand -base64 32
cat /dev/urandom | head -c32 | base64

Set it on your account:


Basic Cucumber feature testing

Cucumber is a Behaviour-Driven-Development (BDD) tool. You can use to automate the acceptance tests for your software systems. A real-world example of its use in a large application is the MediaWiki QA process.

This post will show you the steps to get it up-and-running to test a trivial app on a Unix (Mac, Linux) computer. I’ll assume that you already have Ruby installed.

First up, install Cucumber’s Ruby implementation:

$ sudo gem install cucumber
$ sudo gem install rspec

If you’re on Debian, you can also achieve this with apt-get:

$ sudo apt-get install ruby-cucumber-core ruby-rspec

The versions that I got from the install were-

$ ruby --version
ruby 2.2.3p173 (2015-08-18) [x86_64-linux-gnu]
$ cucumber --version

Make a “hello world” project:

$ mkdir cucumber-hello
$ cd cucumber-hello/

One of the under-celebrated features of Cucumber is the quality of its error messages. Simply attempt to run it and you will be told what to do next:

$ cucumber
No such file or directory - features. You can use `cucumber --init` to get started.

Go ahead and run cucumber --init as the program suggests:

$ cucumber --init
  create   features
  create   features/step_definitions
  create   features/support
  create   features/support/env.rb

This has set up your project. In the same vein as Test-Driven Development (TDD), the usual practice is to write the tests before you code.

Most programmers will be familiar with the expected output of a “Hello World” program. In Cucumber, this ‘feature’ would be represented in Cucumber as:

Feature: Test

  Scenario: Hello
    When I run the "hello-world" program
    Then I should see "Hello World"

Save this definition to features/hello.feature in the project directory and attempt to run Cucumber again:

$ cucumber
Feature: Test

  Scenario: Hello                        # features/hello.feature:3
    When I run the "hello-world" program # features/hello.feature:4
    Then I should see "Hello World"      # features/hello.feature:5

1 scenario (1 undefined)
2 steps (2 undefined)

You can implement step definitions for undefined steps with these snippets:

When(/^I run the "([^"]*)" program$/) do |arg1|
  pending # Write code here that turns the phrase above into concrete actions

Then(/^I should see "([^"]*)"$/) do |arg1|
  pending # Write code here that turns the phrase above into concrete actions

The output helpfully contains the skeleton code for writing the test steps in Ruby. Using this as a base, I implemented this using the rspec expectations syntax, to check if the string starts with “Hello World”.

When(/^I run the "([^"]*)" program$/) do |cmd|
  @value = `./#{cmd}`  

Then(/^I should see "([^"]*)"$/) do |val|
  expect(@value).to start_with(val)

Save these steps to features/step_definitions/hello_steps.rb

Finally, its time to code. Write up an implementation of “Hello World” in basically any available programming language, and save the output to an executable file called hello-world:

Using bash:

echo "Hello World"
$ chmod +x hello-world

Manually testing:

$ ./hello-world
Hello World

Automatically testing with Cucumber confirms that the test criteria are met:

$ cucumber
Feature: Test

  Scenario: Hello                        # features/hello.feature:3
    When I run the "hello-world" program # features/step_definitions/hello_steps.rb:1
    Then I should see "Hello World"      # features/step_definitions/hello_steps.rb:5

1 scenario (1 passed)
2 steps (2 passed)

Now that you’ve completed these steps, you should have this project structure:

$ find .

Next steps

Our scenario for the Hello World program has users interfacing with the system entirely with non-interactive shell commands. Cucumber needs to be able to observe the external behaviour of your application, so your steps will end up including some powerful test libraries. Some that are used often are:

How to arrange pages for printing and cutting in LaTeX

This post covers some LaTeX examples that I recently assembled for print-outs where multiple pages-per-sheet will be printed for cutting.

First up, an example file. This is a document filled with random text, on C6 paper.



\section*{Lorem Ipsum}

It renders like so (click for PDF):


1×1 arrangement

Using the crop package, you can wrap the output pages of your document in larger pages. Using the \includepdf macro, these pages can be sourced from our example document.

This example simply renders the document on larger pages, with crop marks displayed at the document edges.

% Sum of dimensions of the pages. We are displaying 1x1 C6 (114 mm x 162mm) pages, which is where 114x162 comes from.

% Include print/output paper size and orientation here, A4 portrait in this example (use 'landscape' option for landscape)


% Restrict page range, and set arrangement of pages per sheet here.


This renders a single C6 page on A4 like this (click image for PDF):


2×1 arrangement

To lay out two images side-by-side, we need to specify the size of the document we’re wrapping (which will be twice as wide as a single C6 page, and the height of a single C6 page), and also change the orientation of the outer paper to Landscape.

The LaTeX code is:

% Sum of dimensions of the pages. We are displaying 2x1 C6 (114 mm x 162mm) pages, which is where 228x162 comes from.

% Include print/output paper size and orientation here, A4 landscape in this example.


% Restrict page range, and set arrangement of pages per sheet here.


Which renders as (click for PDF):


2×2 arrangement

In a similar way, we can slot four portrait C6 pages on to a piece of A4 portait paper like so:

% Sum of dimensions of the pages. We are displaying 2x2 C6 (114 mm x 162mm) pages, which is where 228x324 comes from.

% Include print/output paper size and orientation here, A3 portrait in this example (use 'landscape' option for landscape)


% Restrict page range, and set arrangement of pages per sheet here.


This renders as:


More complex arrangement

Ok, so all of the above setups had one thing in common: 2 or fewer items on each side, meaning that the half-way axis lines could be used to show where each page ends.

If you are printing many small items, you might fit more than two in a direction, and need to add more axis lines so that you know where to cut!

Here is an example PDF which creates tiny name tags:

% Custom page size

% Macro for a name tag


% Make some name tags for a conference
% Names generated via
\tag{Flynn Craig}{ExampleCorp}
\tag{Daniel Cash}{ExampleCorp}
\tag{Griffith Durham}{ExampleCorp}
\tag{Sean Washington}{ExampleCorp}
\tag{James Simon}{ExampleCorp}
\tag{Aidan Cotton}{ExampleCorp}
\tag{Wing Sanders}{ExampleCorp}
\tag{Dean Wilcox}{ExampleCorp}
\tag{Jerome Flynn}{FooCorp}
\tag{Leonard Steele}{FooCorp}
\tag{Zane Pratt}{FooCorp}
\tag{Andrew Hartman}{FooCorp}
\tag{Roth Mccullough}{FooCorp}
\tag{Fritz Taylor}{Quux Ltd}
\tag{Beau Bray}{Quux Ltd}
\tag{Duncan Hampton}{Quux Ltd}
\tag{Daquan Witt}{Quux Ltd}
\tag{Wylie Heath}{Quux Ltd}
\tag{Thaddeus Payne}{Quux Ltd}
\tag{Carson Ayala}{Quux Ltd}
\tag{Oleg Olson}{Quux Ltd}
\tag{Herrod Gillespie}{Bar University}
\tag{Jonas Wyatt}{Bar University}
\tag{Mannix Patrick}{Bar University}
\tag{Cain Ferguson}{Bar University}
\tag{Nasim Mendez}{Bar University}
\tag{Emery Bennett}{Bar University}
\tag{Oscar Bennett}{Bar University}
\tag{William Neal}{Bar University}
\tag{Dieter Cochran}{Bar University}
\tag{Bert Schultz}{Bar University}


The above example renders like this (click image for PDF):


Now, the trick is that crop is hard-coded to show axis lines only at the half-way points, so take a copy of crop.sty called crop-3x7.sty and find this block:


The block uses \vfill and \hfill to make even spaces.

Edit the file to include more axis lines, like so:

%% 'crop-3x7' is not a real package for general use, it is an example only!
%% It is a modified version of the 'crop' package, for rendering
%% additonal axis marks as cutting aids.
% Changed package name here-
\ProvidesPackage{crop-3x7}[2016/01/03 v2.0 crop marks 3x7 (mf)]
    % Changed from one axis on each side to 2 axes on top/lower edge, 6 on each side, for printing 3x7 panels per this example. 

Now simply prepare a page which shows a 3×7 grid of tags, using the same rules as before, and our modified crop-3x7.sty:

% This file demonstrates the use of a customisation to the 'crop' package to render additional axes for cutting up small name labels.

% Sum of dimensions of the pages. We are displaying 3x7 small tags (60 x 30 mm) to a page, which is where 180x210 comes from.

% Include print/output paper size and orientation here, A3 portrait in this example (use 'landscape' option for landscape)


% Restrict page range, and set arrangement of pages per sheet here.


Render the document, and see the small dashes on the sides which will show you where to cut:


The code

Full source for all of the linked files can be found in my tex-examples repository on GitHub.

How to print red/black on an impact receipt printer

I recently deployed an Epson TM-U220 impact receipt printer. These printers work by striking a ribbon onto the paper, like a type-writer. One of the up-sides to using these intead of a thermal printer is the ability to install a red/black ribbon in place of the default (black) one:



I connected up my printer using a USB-parallel cable, so my previous posts (Linux, Windows) apply for the connector setup.

Using the escpos-php driver on GitHub, a line of red text is printed like this:

 * Example of two-color printing, tested on an epson TM-U220 with two-color ribbon installed.
require_once (dirname ( __FILE__ ) . "/vendor/Escpos.php");

$connector = new FilePrintConnector("/dev/usb/lp0");
try {
    $printer = new Escpos($connector);
    $printer -> text("Hello World!\n");
    $printer -> setColor(Escpos::COLOR_2);
    $printer -> text("Red?!\n");
    $printer -> setColor(Escpos::COLOR_1);
    $printer -> text("Default color again?!\n");
    $printer -> cut();
} finally {
    /* Always close the printer! */
    $printer -> close();

With this result:


How to connect a USB receipt printer up on Mac OS X

This post will show you how to set up a USB receipt printer on Max OS X. These steps were written on Yosemite, but should work on 10.6 onwards (ie, also Snow Leopard through to El Capitan).

This is another post in a series, which has so far covered direct USB printing on Windows and Linux. The printer tested here is this Epson TM-T20:


CUPS is the printing system that’s used on Mac, but most users would be more familiar with the system print dialog:


In our case, we need to set up the printer via the CUPS web interface. This is accessed via a web browser at this address:


At first, you will get knocked back:


To fix this up, open up Applications → Utilities → Terminal and type in:

cupsctl WebInterface=yes

You can then reload the browser and click through to Administration:


Click Add Printer and log in:



Select the USB printer from the list, and optionally share it:



Click Select Another Make/Manufacturer, and select Raw → Raw Queue:




Use the defaults for the other options:


Test print

Type some junk into a file called foo.txt and attempt to print it, using the CUPS printer name:

nano foo.txt
lpr -o raw -H localhost -P EPSON_TM-T20 foo.txt

The prints will be delayed for a few moments, as CUPS spools the jobs.

Disable CUPS web

Once you’re done, for security reasons you should reset this option from before, to disable the web interface to CUPS:

cupsctl WebInterface=no

How to set up sudo on Debian GNU/Linux

If you just installed Debian, you might notice that one Linux staple, sudo is not installed by default. Here is the quick way to add it, and set yourself up as an administrative user.

1. Install sudo, using the su command to elevate privileges.

bob$ su
root# apt-get install sudo

2. Make yourself an administrator. The default /etc/sudoers file contains this line:

# Allow members of group sudo to execute any command

So simply add your account (‘bob‘ in this example) to the sudo group:

root# usermod -a -G sudo bob

3. Log out and back in again, in order to apply changes to your group membership.

bob$ groups
bob cdrom floppy sudo audio dip video plugdev netdev lpadmin scanner bluetooth
bob$ sudo echo test
[sudo] password for bob: 

Fix merge conflicts in git with Meld

When you’re writing code collaboratively, there’s plenty of situations when you need to combine two sets of changes.

This could happen, for example, if Bob and yourself both fix different bugs by making edits to the same file.


This post assumes that your source code is tracked in git.

First up, install meld. The Meld homepage has instructions for other platforms, but on Debian/Ubuntu, it’s just:

sudo apt-get install meld

Now tell git to use it as a tool:

git config --global merge.tool meld

Once you have a merge conflict, you can then fire up Meld like this:

git mergetool

For each file, you will get a 3-way diff. Click the arrows on the sides to move the code you want into the middle:


Once you’ve saved the file and closed Meld, you will be prompted on the command-line. You just tell it whether you’ve successfully merged the file, until it stops giving you new files to merge.

After this, commit the changes:

git commit


Simpler usage

If you don’t use git, you can simply call Meld from the command-line as well. This shows you differences between files in a similar window, and lets you move blocks of code around as well:

meld foo.c bar.c

Continuous testing in PHP with Eclipse

If you already write test cases over your code, then continuous testing is a simple idea: Get fast feedback on your code by running your project’s unit tests very frequently. From PHP, the de facto standard testing tool is PHPUnit, so running tests usually involves typing phpunit into a terminal.

This post will show you how to set up a project with the Eclipse MakeGood plugin, which is a continuous test runner for the Eclipse IDE. When it is enabled, the plugin runs a set of tests every time you save a file.

This is best used on projects which are:

  • Small — so you tests run in a reasonable timeframe.
  • Complex — reasonably high risk of introducing defects.
  • Well-tested — so your unit tests are reasonably likely to identify defects.

Set up your machine

Your machine needs Eclipse PDT, MakeGood and composer.

Eclipse PDT

If you’ve found this post via Google, then you probably already run Eclipse and PHP. If not, fetch the “Eclipse for PHP developers” distribution from here

You need to pick up the XDebug extension to really make use of this setup. On Debian-based distributions, this is achieved via:

sudo apt-get install php5-xdebug

If you don’t use Eclipse yet, then check whether languages or tools you’ve used before have plugins in the Eclipse Marketplace. There’s a good chance that Eclipse will do the whole lot.


MakeGood is simply an Eclipse plugin, so you can fetch it from the Eclipse Marketplace.

Navigate to Help → Eclipse Marketplace and search for “makegood”:


Wait for the info to be fetched:

Check all the boxes, and then click Confirm:


Accept the terms, and wait for the install:


At one point, you will be prompted to accept a developer’s certificate, which you need to do before continuing:


This feature will not work unless you have a PHP binary set up. You need to configure this under Window → Preferences → PHP → PHP executables.:


This is system dependent: Windows users will find a file somewhere called php-cli.exe, while Linux users might find it at /usr/bin/php.

After this, MakeGood is installed. It is configured per-project, so we’ll get to that after the next section.


Composer is a dependency manager which is widely used in the PHP world.

You can get it from

Rather than use the install process which is suggested there, you can simply download the phar, make it executable, and place it somewhere like /usr/bin:

chmod +x composer.phar
sudo mv composer.phar /usr/bin/composer
composer --version

By using composer to load PHPUnit, MakeGood becomes much easier to configure.

Project setup

To start, make a new PHP project by clicking File → New → PHP Project. Name it something like “Example”.

We are using composer here because it provides an easy way to make sure MakeGood can see the PHPUnit code. So let’s set a few things up:

First, add a composer.json file which adds a recent version of PHPUnit as a require-dev dependency. Right-click Example project, and navigate to New → File, naming it composer.json:

    "require-dev": {
        "phpunit/phpunit": "4.5.*"

Note that with the way that MakeGood constructs its command-line, newer versions of PHPUnit do not appear to work correctly at the time of writing.

Next up, install the new dependency with the help of Composer:

$ composer install
Loading composer repositories with package information
Installing dependencies (including require-dev)
  - Installing sebastian/version (1.0.6)
  - Installing sebastian/global-state (1.0.0)
  - Installing sebastian/recursion-context (1.0.1)
  - Installing sebastian/exporter (1.2.1)
  - Installing sebastian/environment (1.3.2)
  - Installing sebastian/diff (1.3.0)
  - Installing sebastian/comparator (1.2.0)
  - Installing symfony/yaml (v2.7.3)
  - Installing doctrine/instantiator (1.0.5)
  - Installing phpdocumentor/reflection-docblock (2.0.4)
  - Installing phpspec/prophecy (v1.4.1)
  - Installing phpunit/php-text-template (1.2.1)
  - Installing phpunit/phpunit-mock-objects (2.3.6)
  - Installing phpunit/php-timer (1.0.7)
  - Installing phpunit/php-token-stream (1.4.3)
  - Installing phpunit/php-file-iterator (1.3.4)
  - Installing phpunit/php-code-coverage (2.2.2)
  - Installing phpunit/phpunit (4.5.1)

sebastian/global-state suggests installing ext-uopz (*)
phpdocumentor/reflection-docblock suggests installing dflydev/markdown (~1.0)
phpdocumentor/reflection-docblock suggests installing erusev/parsedown (~1.0)
phpunit/phpunit suggests installing phpunit/php-invoker (~1.1)
Writing lock file
Generating autoload files

Set up an Example.php class with a bit of code:

 * Example of some code that could do with test cases
class Example {
	 * Faulty palindrome checker. A palindrome is a word which is the same when read orward and backward.
	 * @param string $str
	 * @return boolean True if $str is a palindrome, false if it is not.
	function isPalindrone($str) {
		$halfLen = (int)(strlen($str) / 2);
		for($i = 0; $i < $halfLen; $i++) {
			// Compare characters from the front and back of the string, and move toward the middle.
			$c1 = substr($str, $i, 1);
			$c2 = substr($str, (2 * $halfLen - 1) - $i, 1);
			if($c1 != $c2) {
				return false;
		return true;


Set up a test/unit folder and create a ExampleTest class with a useful test case in it, like so:

class ExampleTest extends PHPUnit_Framework_TestCase {
	public function testIsPalindrome() {
		$e = new Example();
		$this -> assertTrue($e -> isPalindrone(""));
		$this -> assertTrue($e -> isPalindrone("a"));
		$this -> assertTrue($e -> isPalindrone("ABBA"));
		$this -> assertFalse($e -> isPalindrone("test"));

The unit tests go in their own folder, in case you decide to add more extensive (slower) tests later. MakeGood should only be configured to run fast tests!

Now add a test/bootstrap.php for test cases to load classes (NB: At scale, you would have composer autoload your classes, and use that as your PHPUnit class loader).

require_once 'vendor/autoload.php';
require_once 'Example.php';

Optionally, also add a phpunit.xml file in your project root to configure PHPUnit with, so that you can run it on the command-line:


If all goes to plan, your project should look something like this:

- Continuous testing example
  |- (.. code ..)
  |- Example.php
  |- test/
  |  |- (.. other types of tests ..) 
  |  |- unit/
  |  | |- ExampleTest.php
  |  |- bootstrap.php
  |- vendor
  |  |- (.. dependencies ..)
  |  |- autoload.php
  |- composer.json
  |- phpunit.xml

And more to the point, you should be able to use PHPUnit on the command-line to run the test case. It will look something like this:

Example$ php vendor/bin/phpunit test/unit/
PHPUnit 4.7.7 by Sebastian Bergmann and contributors.


Time: 71 ms, Memory: 4.50Mb

OK (1 test, 4 assertions)

Configure MakeGood

Now that the project is coming along nicely, it’s time to configure MakeGood to run the test automatically from within Eclipse.

Go to Project → Properties → MakeGood, and add the path to your unit tests, along with a link to the bootstrap.php. This bootstrap file will let MakeGood load the classes to test, along with PHPUnit itself.


Now fire up the MakeGood view. If it doesn’t come up automatically, then open it via Window → Show View → Other. Clicking the “play” button should kick off all available tests, adn if your project structure is correct, then everything will work at this point:


Once you save a file, a small box will appear in the bottom-right to indicate whether the tests worked.


As you introduce changes and save them, MakeGood will flag whether the tests are passing by changing the colour of the icon.

In this case, it is green. From the small delay that is caused when running the tests, you can see why it’s important to keep unit tests fast if you use this tool.

A test-driven bugfix cycle

Of course, Linus’ Law suggests that any code will probably have more bugs. So, say somebody reports that the palindrome “abcdedcba” is causing incorrect results. In test-driven development, you would expose this as a test case over the isPalindrome() function:

	public function testIsPalindromeOddLength() {
		$e = new Example();
		$this -> assertTrue($e -> isPalindrone("abcdedcba"));
		$this -> assertFalse($e -> isPalindrone("abc"));

As soon as this is submitted, things start to turn red in MakeGood. In response to the broken test, you can then tweak the code.


You would then do the smallest amount of programming needed to pass all the tests. A palindrome checker which does not contain intentional bugs is as simple as this one-liner:

	function isPalindrone($str) {
		return $str == strrev($str);

This change causes MakeGood to report that the unit tests are all passing.



A continuous test runner adds faster feedback for tedious but well-tested code. If you have to write anything in PHP which does some real computer work, then I suggest setting up MakeGood so that you can make changes to tricky code, and have a good idea of whether it’s producing sane outputs.

As this post has touched on before, slow tests should not run each time you save the code. This is why we worked in a Unit test folder. Slower integration or functional tests can still be run via the command-line in this model. An example of a project which uses this setup is escpos-php, which is a heavily test-driven PHP driver for thermal line printers.

It’s time to migrate away from Outlook Express

Outlook Express is obsolete, so you need to migrate if you’re still using it. This post is a quick guide to saving your local data so that you can jump ship.

If you are keen on desktop-based email, then there are only two real contenders for a replacement mail client:

  • Mozilla Thunderbird (suggested).
  • Windows Mail.

So where is all my data?

The script below is a Windows bat script for backing up an Outlook Express setup. Just fill in the Identity variable and run it from anywhere to produce a folder containing the Outlook saved emails and contacts.

You can find your identity string as a folder name in your profile path, under Local Settings\Application Data\Identities\. The profile path for a user called bob would usually be in a folder like C:\Documents and Settings\bob or C:\Users\bob.

These files can be read by both of the suggested replacements, so a good transition might be:

  1. Make a backup.
  2. Install an alternative & set it up.
  3. Try to import as much as you can.
  4. Once you’re happy, uninstall Outlook Express, or at least delete the shortcuts to it.
@echo off
ECHO -------------------------------------------------------------------------------
SET BACKUPDIR="%username% - Outlook Backup"
ECHO Backing up outlook files to: %backupdir%
ECHO --------------------------------------------------------------------------------

ECHO - Clearing backup location... (1 of 3) 
DEL /F /S /Q %backupdir% > NUL 2>&1
MKDIR %backupdir% 2> NUL
ECHO - - Done

ECHO - Copying emails... (2 of 3) 
MKDIR %backupdir%\Emails 2> NUL
XCOPY /E /H /C /R /Y "%userprofile%\Local Settings\Application Data\Identities\%IDENTITY%\Microsoft\Outlook Express" %backupdir%\Emails > NUL
ECHO - - Done

ECHO - Copying address book... (3 of 3)
MKDIR %backupdir%\AddressBook 2> NUL
XCOPY /E /H /C /R /Y "%appdata%\Microsoft\Address Book" %backupdir%\AddressBook > NUL
ECHO - - Done

ECHO - Completed 3 of 3 tasks.
ECHO -------------------------------------------------------------------------------
ECHO The backup is complete.
ECHO Please copy %BACKUPDIR% to external storage.
ECHO --------------------------------------------------------------------------------

Raspberry Pi KA Lite wireless deployment

Previous post in this series: How to install KA Lite on the Raspberry Pi

If you got to the end of the previous post, you would have a working install of the open source KA Lite system on the Raspberrry Pi.

This post will show you how to deploy this KA Lite setup without relying on any existing network infrastructure:


To share the setup without an existing network, we will configure the wireless adapter to provide its own network for computers to connect to. If you have an existing wireless network, you can skip this post.

Before you begin

This post is intended for the widely-used “WiPi” USB WiFi adapter, so you’ll need one of those.

I used the same Tenda USB WiFi adapter as previous setups, which has a Ralink chip from the same series:

USB WiFi Adapter

Set it up

See the official guide for these steps in compact form. These steps are the same, but expanded with extra output from my install.

Clone the raspberry pi scripts repo:

$ git clone
Cloning into 'ka-lite-pi-scripts'...
remote: Counting objects: 66, done.
remote: Total 66 (delta 0), reused 0 (delta 0), pack-reused 66
Unpacking objects: 100% (66/66), done.

Open it up and run ./ as the root user. This requires Internet access, and will install and configure a few packages:

$ cd ka-lite-pi-scripts/
$ ls
$ sudo ./
Get:1 wheezy Release.gpg [490 B] 
Hit wheezy Release.gpg                        
Hit wheezy Release.gpg
Get:2 wheezy Release [14.4 kB]
Hit wheezy Release                      
Hit wheezy Release                        
Hit wheezy/rpi armhf Packages           
Get:3 wheezy/main armhf Packages [129 kB]       
Get:4 wheezy/main armhf Packages [6903 kB]  
Ign wheezy/rpi Translation-en                 
Ign wheezy/main Translation-en                
Get:5 wheezy/contrib armhf Packages [23.6 kB]
Get:6 wheezy/non-free armhf Packages [49.3 kB]
Get:7 wheezy/rpi armhf Packages [592 B]     
Ign wheezy/contrib Translation-en           
Ign wheezy/main Translation-en              
Ign wheezy/non-free Translation-en          
Ign wheezy/rpi Translation-en
Fetched 6992 kB in 35s (198 kB/s)             
Reading package lists... Done
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following packages will be upgraded:
  cups-bsd cups-client cups-common fuse libcups2 libcupsimage2 libfuse2 raspi-config
8 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
Need to get 1708 kB of archives.
After this operation, 306 kB disk space will be freed.
Get:1 wheezy/main raspi-config all 20150131-4 [13.2 kB]
Get:2 wheezy/main libcupsimage2 armhf 1.5.3-5+deb7u6 [132 kB]
Get:3 wheezy/main cups-common all 1.5.3-5+deb7u6 [904 kB]
Get:4 wheezy/main cups-bsd armhf 1.5.3-5+deb7u6 [44.2 kB]
Get:5 wheezy/main cups-client armhf 1.5.3-5+deb7u6 [174 kB]
Get:6 wheezy/main libcups2 armhf 1.5.3-5+deb7u6 [238 kB]
Get:7 wheezy/main fuse armhf 2.9.0-2+deb7u2 [70.8 kB]
Get:8 wheezy/main libfuse2 armhf 2.9.0-2+deb7u2 [132 kB]
Fetched 1708 kB in 3s (447 kB/s)
Preconfiguring packages ...
(Reading database ... 83674 files and directories currently installed.)
Preparing to replace libcupsimage2:armhf 1.5.3-5+deb7u5 (using .../libcupsimage2_1.5.3-5+deb7u6_armhf.deb) ...
Unpacking replacement libcupsimage2:armhf ...
Preparing to replace cups-common 1.5.3-5+deb7u5 (using .../cups-common_1.5.3-5+deb7u6_all.deb) ...
Unpacking replacement cups-common ...
Preparing to replace cups-bsd 1.5.3-5+deb7u5 (using .../cups-bsd_1.5.3-5+deb7u6_armhf.deb) ...
locale: Cannot set LC_ALL to default locale: No such file or directory
Unpacking replacement cups-bsd ...
Preparing to replace cups-client 1.5.3-5+deb7u5 (using .../cups-client_1.5.3-5+deb7u6_armhf.deb) ...
Unpacking replacement cups-client ...
Preparing to replace libcups2:armhf 1.5.3-5+deb7u5 (using .../libcups2_1.5.3-5+deb7u6_armhf.deb) ...
Unpacking replacement libcups2:armhf ...
Preparing to replace fuse 2.9.0-2+deb7u1 (using .../fuse_2.9.0-2+deb7u2_armhf.deb) ...
Unpacking replacement fuse ...
Preparing to replace libfuse2:armhf 2.9.0-2+deb7u1 (using .../libfuse2_2.9.0-2+deb7u2_armhf.deb) ...
Unpacking replacement libfuse2:armhf ...
Preparing to replace raspi-config 20150131-3 (using .../raspi-config_20150131-4_all.deb) ...
Unpacking replacement raspi-config ...
Processing triggers for man-db ...
Processing triggers for initramfs-tools ...
Setting up libcups2:armhf (1.5.3-5+deb7u6) ...
Setting up libcupsimage2:armhf (1.5.3-5+deb7u6) ...
Setting up cups-common (1.5.3-5+deb7u6) ...
Setting up cups-client (1.5.3-5+deb7u6) ...
Setting up cups-bsd (1.5.3-5+deb7u6) ...
Setting up libfuse2:armhf (2.9.0-2+deb7u2) ...
Setting up fuse (2.9.0-2+deb7u2) ...
udev active, skipping device node creation.
update-initramfs: deferring update (trigger activated)
Setting up raspi-config (20150131-4) ...
update-rc.d: warning: default start runlevel arguments (2 3 4 5) do not match raspi-config Default-Start values (S)
update-rc.d: warning: default stop runlevel arguments (0 1 6) do not match raspi-config Default-Stop values (none)
Processing triggers for initramfs-tools ...
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following extra packages will be installed:
  dnsmasq-base libnetfilter-conntrack3
The following NEW packages will be installed:
  dnsmasq dnsmasq-base hostapd libnetfilter-conntrack3
0 upgraded, 4 newly installed, 0 to remove and 0 not upgraded.
Need to get 822 kB of archives.
After this operation, 1653 kB of additional disk space will be used.
Get:1 wheezy/main libnetfilter-conntrack3 armhf 1.0.1-1 [32.2 kB]
Get:2 wheezy/main dnsmasq-base armhf 2.62-3+deb7u3 [356 kB]
Get:3 wheezy/main dnsmasq all 2.62-3+deb7u3 [16.3 kB]
Get:4 wheezy/main hostapd armhf 1:1.0-3+deb7u2 [418 kB]
Fetched 822 kB in 4s (168 kB/s)  
Selecting previously unselected package libnetfilter-conntrack3:armhf.
(Reading database ... 83674 files and directories currently installed.)
Unpacking libnetfilter-conntrack3:armhf (from .../libnetfilter-conntrack3_1.0.1-1_armhf.deb) ...
Selecting previously unselected package dnsmasq-base.
Unpacking dnsmasq-base (from .../dnsmasq-base_2.62-3+deb7u3_armhf.deb) ...
Selecting previously unselected package dnsmasq.
Unpacking dnsmasq (from .../dnsmasq_2.62-3+deb7u3_all.deb) ...
Selecting previously unselected package hostapd.
Unpacking hostapd (from .../hostapd_1%3a1.0-3+deb7u2_armhf.deb) ...
Processing triggers for man-db ...
Setting up libnetfilter-conntrack3:armhf (1.0.1-1) ...
Setting up dnsmasq-base (2.62-3+deb7u3) ...
Setting up dnsmasq (2.62-3+deb7u3) ...
[ ok ] Starting DNS forwarder and DHCP server: dnsmasq.
Setting up hostapd (1:1.0-3+deb7u2) ...
[ ok ] Starting advanced IEEE 802.11 management: hostapd.
[ ok ] Starting DNS forwarder and DHCP server: dnsmasq[....] (already running).
update-rc.d: using dependency based boot sequencing
update-rc.d: using dependency based boot sequencing

Next, run run the ./ script:

$ ./ 
[ ok ] Stopping advanced IEEE 802.11 management: hostapd.
[ ok ] Starting advanced IEEE 802.11 management: hostapd.

And then the ./ script as root:

sudo python ./

After this ran, the /etc/network/interfaces file looked like:

$ sudo nano /etc/network/interfaces
auto lo
iface lo inet loopback

auto eth0
#allow-hotplug eth0
iface eth0 inet manual

auto wlan0
#allow-hotplug wlan0
wpa-conf /etc/wpa_supplicant/wpa_supplicant.conf

auto wlan1
#allow-hotplug wlan1
iface wlan1 inet manual
wpa-conf /etc/wpa_supplicant/wpa_supplicant.conf

iface wlan0 inet static
  pre-up sudo iptables -t nat -A PREROUTING -i wlan0 -p tcp -j DNAT --to-destination
  pre-up sudo python /home/pi/ka-lite/ka-lite-pi-scripts/ &

And I modified /etc/default/ifplugd to only include eth0, so that the pi does not attempt to join any wireless networks itself:

$ sudo nano /etc/default/ifplugd
ARGS="-q -f -u0 -d10 -w -I"

And then reboot:

$ sudo reboot

Broadcast message from root@raspberrypi (pts/0) (Thu Jun 11 11:05:46 2015):

The system is going down for reboot NOW!

See if its all working

Disregard the old addresses once you’ve done the above steps, and run ifconfig to find out about addresses after rebooting.

The pi broadcasts a passwordless wireless network called “kalite”, and all DNS lookups point to

This means in a web browser, you can just load a URL like:



On a laptop in range, the network appeared once the pi booted:


The network connected without a password:


And the connection info for the client computer is being checked in the background:


Once the computer gets a lease from DHCP, the URL loads, and the interface works as expected:


Tested on Android, the player and navigation worked in a similar way:


KA Lite’s media player doesn’t seem to require anything other than a web browser. Either Chrome or Firefox seem to be suitable choices.

Debug zone

Some useful commands for debugging are shown below.

List USB devices, for debugging the wireless adapter:


List recent errors, for debugging the same thing:


Restart networking, run when changing configuration:

service networking restart
Note on dnsmasq

Because of the way dnsmasq hijacks lookups to provide the wireless network, the Raspberry pi cannot access the Internet over Ethernet (eg, to download content) while dnsmasq is running:

sudo service dnsmasq stop

You can access the web interface at while this service is disabled.

And then start it again when done using the Internet, so that the user-friendly computer name works again:

sudo service dnsmasq start