Personal Website of David Taylor on Cloud Artisan

Blue Mountains Taekwondo

Mon, 01 Jan 0001 00:00:00 +0000

A martial arts school teaching traditional Taekwondo in the Blue Mountains, NSW.

Palette Kit

Mon, 01 Jan 0001 00:00:00 +0000

A digital tool to help artists find the perfect coloured pencil combinations for their artwork.

Shrnk.ai

Mon, 01 Jan 0001 00:00:00 +0000

A modern URL shortening service with AI-powered features for link management.

RunMyDojo

Mon, 01 Jan 0001 00:00:00 +0000

A comprehensive management system for martial arts schools and dojos.

Old School TKD

Mon, 01 Jan 0001 00:00:00 +0000

A resource for traditional Taekwondo techniques, forms, and training methods.

Claude Code Tips & Tricks: Working with Lovable

Tue, 02 Sep 2025 00:00:00 +1000

When AI Tools Collide: A Real-World Experience #

While building my latest project, Palette Kit (a React/TypeScript app for managing colored pencil gradients), I discovered something interesting: using Claude Code and Lovable together is incredibly powerful, but it requires a specific approach to avoid some nasty pitfalls.

Honestly, I nearly broke my entire deployment pipeline learning this the hard way. But here’s what I learned - and how you can make these tools work beautifully together without the pain I went through.

Why Change Moratoriums Don't Work

Mon, 04 Aug 2025 21:30:00 +1000

I’ve been in this situation more times than I’d care to count: production is unstable, customers are unhappy, and leadership decides the solution is to freeze all changes until further notice. On the surface, it makes perfect sense. If changes are causing problems, stop making changes, right?

Wrong!

I understand the logic. I understand why executives think they should make this call. They assess the business impact of continuing with potentially risky changes versus grinding everything to a halt.

Essential Books

Mon, 07 Jul 2025 15:20:00 +1000

My Reading Philosophy #

This is my collection of essential books - stories and ideas that have shaped my thinking and sparked my imagination. I’m drawn to speculative fiction that explores profound themes, spy fiction with psychological complexity, and non-fiction that challenges conventional wisdom.

My favourite authors include Ursula K. Le Guin for her philosophical depth, John Wyndham for his “cosy catastrophes”, China Miéville for inventive world-building, Barry Eisler for morally complex thrillers, and David Epstein for thought-provoking analysis of human potential.

Essential Films

Mon, 07 Jul 2025 15:15:00 +1000

My Film Philosophy #

This is my collection of essential films - it’s a blend of favourite movies and a list of films that must be seen. I’m drawn to drama that makes you think and feel deeply, quality sci-fi that isn’t derivative and inspires thought, and martial arts films with compelling storytelling and exceptional choreography.

It’s very rare that I like a popcorn movie, but if it hits all the right points for a bit of vacuous fun without any groan-inducing moments or preachy, moralistic messaging, I’m in.

Getting Started with Google Gemini CLI

Fri, 04 Jul 2025 19:10:04 +1000

Google recently released their own entry into the agentic CLI arena: Gemini CLI. What might set it apart from OpenAI’s Codex CLI and Anthropic’s Claude Code is Gemini’s large context window and support for multimodal input.

I’ve started testing it out to see just how it compares to Codex CLI and Claude Code (the latter of which I’ve been using regularly). This is where I started…

What You Can Do with Gemini CLI #

With Gemini CLI you can:

OpenAI Codex Tips & Tricks: Listing Available Models

Mon, 21 Apr 2025 09:00:00 +1000

After getting started with OpenAI Codex CLI, I’ve discovered some handy tips to make working with it a bit easier.

Listing Available Models #

When working with Codex CLI, you can specify which model to use with the -m flag:

codex -m o4-mini "explain this code"

But what if you want to know all the models available to you? The default documentation doesn’t provide a straightforward way to list them. Here’s a simple one-liner that queries the OpenAI API to list all models available to your account:

Getting Started with OpenAI Codex CLI

Fri, 18 Apr 2025 12:00:00 +1000

I’ve been playing with the recently announced OpenAI Codex CLI.

What is OpenAI Codex CLI? #

OpenAI Codex CLI is available on GitHub. It’s a lightweight AI coding assistant that runs directly in the terminal, like Claude Code from Anthropic.

Features include:

Read, modify, and run code
Multimodal input for processing screenshots and diagrams
Different approval levels

While it’s still labeled as experimental, OpenAI is actively seeking community contributions. They’re even offering a $1 million initiative supporting open-source projects that use Codex CLI, with grants in $25,000 API credit increments.

Claude Code Tips & Tricks: Maximising Memory

Wed, 16 Apr 2025 10:55:08 +1000

The Power of Memory in Claude Code #

Having used Claude Code for a while now to help manage this website and other projects, I’ve come to appreciate how crucial its memory features are for productive work. In AI assistants like Claude, memory refers to the system’s ability to retain information across interactions, which can significantly reduce repetitive instructions and improve continuity.

Claude Code Tips & Tricks: Custom Slash Commands

Mon, 14 Apr 2025 15:18:37 +1000

The Power of Custom Slash Commands #

I’ve been using Claude Code to help manage this website, and I kept finding myself typing similar instructions repeatedly: “Create a new draft post titled <X>,” “Check this file for spelling, grammar, and UK English,” “Clear the cache and start the server” in a very wordy way with a lot of back and forth – these repetitive tasks were begging to be streamlined.

Claude Code Tips & Tricks: Setting Up MCP Servers

Sat, 12 Apr 2025 00:00:00 +1000

What Are MCP Servers? #

Model Context Protocol (MCP) servers extend AI Assistants’ capabilities by connecting them to external tools and services.

Recently, Claude Code gained support for MCP servers, enabling integration with more specialised services like Stripe, Cloudflare, Supabase, Blender, and a whole lot more.

This post demonstrates setting up some simple reference MCP servers. These servers won’t dramatically enhance Claude Code, since it already has built-in tools for filesystem, git, and web fetching. However, they serve as a gentle introduction to installing and configuring MCP servers without having to worry about overly-complex configuration. We can tackle more advanced integrations later in a future post.

Digital Diaspora: Where Egan's 'Diaspora' Meets 'Pantheon'

Mon, 07 Apr 2025 00:00:00 +1000

I’ve been thinking about digital consciousness lately. A lot. What happens when human minds transition into a digital realm?

Two works explore this territory in fascinating ways. Greg Egan’s novel “Diaspora” (1997), which I finally got round to reading a year ago. And the animated series “Pantheon” (2022), which I just finished watching.

“Diaspora” is an ambitious novel set in a posthuman future where consciousness exists primarily as software. Beginning with a catastrophe that threatens Earth, the novel follows digital entities on a journey spanning thousands of years and multiple dimensions of space. It’s a rigorous exploration of what existence might mean when freed from biological constraints. It can be a challenging read at times, but well worth it.

The Turing Option: A Prophetic Vision of Modern AI

Wed, 02 Apr 2025 00:00:00 +1100

Recently, I found myself reminiscing about a sci-fi novel I read in high school: The Turing Option by Marvin Minsky and Harry Harrison. Published in 1992, this book has been on my mind as I’ve watched the recent explosion in development of modern AI.

What stands out isn’t just that it was co-authored by Minsky: one of the founding fathers of artificial intelligence–but how many concepts in the book parallel today’s AI landscape.

Building Palette Kit: From Zero to Production in a Weekend

Mon, 31 Mar 2025 00:00:00 +1100

So I built Palette Kit over the weekend. The whole thing, start to finish: from zero lines of code to a published site.

The Spark #

It began with a phone call with my sister. She runs My Colourful Country Life, a site for adult colouring enthusiasts. As my sister explained the value of her PDFs for artists, I realised it’d translate VERY well into an application with search and other features.

About Me

Sat, 09 Sep 2023 00:00:00 +1000

Cloud Artisan #

I’m David Taylor, a software engineer and cloud architect with a passion for technology, martial arts, and retro computing.

With over 20 years of experience in the tech industry, I’ve worked on various projects ranging from cloud infrastructure to software development. This blog serves as a platform to share my knowledge, experiences, and insights.

Interests #

Cloud Computing and DevOps
Software Development
Retro Computing and Technology
Martial Arts (Taekwondo)
Open Source Projects

Projects #

Check out my projects page to see what I’m currently working on.

Google App Engine for Python: Getting Started On Mac

Mon, 22 Aug 2011 00:00:00 +1000

Download the DMG package from http://code.google.com/appengine/downloads.html and open it.

Drag GoogleAppEngineLauncher to your Applications folder. Press CMD-Space to bring up Spotlight and type in “GoogleAppEngineLauncher” and hit Enter.

Agree to opening the application.

Click on OK when prompted to make command symbolic links.

You’ll be prompted for your password. Once you’ve provided your password the symbolic links should be created; take note of the available commands:

Installing mr.awsome on Mac OS X

Tue, 21 Jun 2011 00:00:00 +1000

To quote the mr.awsome documentation “mr.awsome is a commandline-tool (aws) to manage and control Amazon Webservice’s EC2 instances.”

Once you know the correct incantation, installing it is easy.

curl -o setuptools-0.6c11-py2.6.egg http://pypi.python.org/packages/2.6/s/setuptools/setuptools-0.6c11-py2.6.egg#md5=bfa92100bd772d5a213eedd356d64086
sudo sh setuptools-0.6c11-py2.6.egg
rm setuptools-0.6c11-py2.6.egg
export ARCHFLAGS="-arch i386 -arch x86_64"
easy_install mr.awsome

Simple as that.

Follow me on Twitter. If you don’t, kittens and fairies will die.

Resizing an RDS database instance

Mon, 30 May 2011 00:00:00 +1000

I keep forgetting how to do this, so…

 wintermute:~ $ rds-modify-db-instance mydb --region us-east-1 --db-instance-class db.m2.2xlarge --apply-immediately
 DBINSTANCE mydb 2011-05-20T01:24:19.131Z db.m1.large mysql 250 master available mydb.abcdefghijk.us-east-1.rds.amazonaws.com 3306 us-east-1b 0 db.m2.2xlarge n 5.5.8
 SECGROUP default active
 PARAMGRP mydb-params in-sync

Yup, that involves an immediate outage for that database.

Yup, that’s the really expensive one…

Resizing EC2 Instances

Mon, 30 May 2011 00:00:00 +1000

An EC2 instance needs to be stopped before it can be resized. Also, it can only be resized to a “compatible” type (ie, one with a compatible kernel, 32 bit or 64 bit). So, if you started with a 32 bit t1.micro you can only resize up to a 32 bit m1.small. If you started with a 64 bit t1.micro you could resize up to a 64 bit m1.large.

wintermute:~ $ ec2-stop-instances i-3e15ff51
INSTANCE i-3e15ff51 running stopping
wintermute:~ $ ec2-modify-instance-attribute -t m2.2xlarge i-3e15ff51
Client.IncorrectInstanceState: The instance 'i-3e15ff51' is not in the 'stopped' state.
wintermute:~ $ ec2-modify-instance-attribute -t m2.2xlarge i-3e15ff51
instanceType i-3e15ff51 m2.2xlarge
wintermute:~ $ ec2-start-instances i-3e15ff51
INSTANCE i-3e15ff51 stopped pending

By the way, these short posts are essentially notes to myself while working (I keep forgetting the syntax!). I’ve got some Cherokee-related posts in the pipeline, I just need to find some extra spare time at night.

Server Density Plugin for Monitoring Cherokee

Sat, 07 May 2011 00:00:00 +1000

I’ve created a Server Density plugin that can be used to monitor the Cherokee web server. The plugin is currently available on github. The installation instructions are over there. If you have any changes or improvements you’d like to see, please fork and create pull requests.

Here are some screenshots…

Getting Started With AWS Identity Access Management

Mon, 18 Apr 2011 00:00:00 +1000

AWS Identity Access Management (IAM) is a free feature provided with your AWS account…

enabling businesses to create multiple Users with individual security credentials who can use AWS web services, all controlled by and billed to a single AWS Account

For example, you might use it to:

restrict your web application to only manage S3
give accounts to an operations team, adding them to an operations group that can reboot EC2 instances
bring on contractors for short-term work
prevent yourself from accidentally stopping or deleting your RDS instance
enforce separation of your testing, staging, and production environments
and much more

How many of you are using the credentials of your AWS ‘master’ account to do everything? Well… simply put… don’t do that!

Installing newrelic_api gem from GitHub

Mon, 07 Mar 2011 00:00:00 +1100

I needed to correlate some logs with New Relic monitoring data. I started doing this manually… and got bored very quickly. It was too slow and painstaking. After some poking around I came across newrelic_api (not to be confused with rpm, which is the monitoring agent).

Next thing I discovered… I had no idea how to install gems from GitHub!

No longer!

wintermute:~/Git $ mkdir newrelic
wintermute:~/Git $ cd newrelic/
wintermute:~/Git/newrelic $ git clone git://github.com/newrelic/newrelic_api.git
Cloning into newrelic_api...
remote: Counting objects: 132, done.
remote: Compressing objects: 100% (116/116), done.
remote: Total 132 (delta 59), reused 0 (delta 0) Receiving objects: 100% (132/132), 26.02 KiB, done.
Resolving deltas: 100% (59/59), done.
wintermute:~/Git/newrelic $ cd newrelic_api/
wintermute:~/Git/newrelic/newrelic_api (master)$ ls
CHANGELOG Gemfile.lock README.rdoc VERSION log test Gemfile LICENSE.txt Rakefile lib newrelic_api.gemspec
wintermute:~/Git/newrelic/newrelic_api (master)$ gem build newrelic_api.gemspec
Successfully built RubyGem
Name: newrelic_api
Version: 1.1.1
File: newrelic_api-1.1.1.gem
wintermute:~/Git/newrelic/newrelic_api (master)$ gem install newrelic_api-1.1.1.gem
Successfully installed newrelic_api-1.1.1
1 gem installed
Installing ri documentation for newrelic_api-1.1.1...
Installing RDoc documentation for newrelic_api-1.1.1...
wintermute:~/Git/newrelic/newrelic_api (master)$

It’s as simple as that.

Install Ruby 1.9.2 and Rails 3.0.3 using rvm on Ubuntu 10.10

Fri, 10 Dec 2010 00:00:00 +1100

It’s terse, probably more for my benefit than yours, but let’s dive in…

Install the packages required by rvm:

david@continuity:~$ sudo apt-get install curl git

Install rvm:

david@continuity:~$ bash < <( curl http://rvm.beginrescueend.com/releases/rvm-install-head )
% Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed
100 986 100 986 0 0 524 0 0:00:01 0:00:01 --:--:-- 4522
Initialized empty Git repository in /home/david/.rvm/src/rvm/.git/
remote: Counting objects: 16240, done.
remote: Compressing objects: 100% (4166/4166), done.
remote: Total 16240 (delta 10951), reused 15861 (delta 10649)
Receiving objects: 100% (16240/16240), 2.91 MiB | 457 KiB/s, done.
Resolving deltas: 100% (10951/10951), done.

RVM: Shell scripts enabling management of multiple ruby environments.
RTFM: http://rvm.beginrescueend.com/
HELP: http://webchat.freenode.net/?channels=rvm (#rvm on irc.freenode.net)

Installing RVM to /home/david/.rvm/
Correct permissions for base binaries in /home/david/.rvm/bin...
Copying manpages into place.

Backup and update your .bashrc to make sure rvm is ready whenever you fire up a terminal:

Upgrading Djangy

Wed, 08 Dec 2010 00:00:00 +1100

With the newer versions of Djangy comes support for the fantastic pip installer.

It’s as easy as:

david@continuity:~$ sudo pip install --upgrade djangy
Downloading/unpacking djangy
Downloading Djangy-0.14.tar.gz
Running setup.py egg_info for package djangy
Installing collected packages: djangy
Found existing installation: Djangy 0.11
Uninstalling Djangy: Successfully uninstalled Djangy
Running setup.py install for djangy
Installing djangy script to /usr/local/bin
Successfully installed djangy
Cleaning up...

My first Djangy project (local testing)

Sat, 04 Dec 2010 00:00:00 +1100

In my previous post I knocked out some simple code and pushed it straight to Djangy. I didn’t even test it locally first (gosh shock horror aghast)! Well, that must end… now.

A Djangy project is, in it’s heart of hearts, a Django project that you’ve shoved out your door into the big bad world. And the great thing about Django projects… you can run them locally for testing.

Simply go to your project’s directory. In there you’ll find manage.py. This handy script will do a lot of project management tasks, but the main one I’m interested in is running a local version of my project.

My first Djangy project (initial testing)

Thu, 02 Dec 2010 00:00:00 +1100

My first Djangy project will be an application called rightscalefeed. It will pull down a user’s RightScale event feed. Ideally, I’d like to transform the feed into a WebSocket, suitable for continuous updates, perhaps for display in a data centre or network operations centre.

I may have bitten off more than I can chew, as I’m not sure Django (or Djangy’s environment) has WebSocket support yet, but we’ll see…

First, I need Djangy:

Tuning PHP APC with Cherokee - Round 1

Mon, 29 Nov 2010 00:00:00 +1100

Since writing Using PHP APC with Cherokee I noticed that my cache hits were dropping and my cache misses were growing. This is my first attempt at tweaking the configuration of php-apc to try to eke out more performance.

I enabled my apc.php page (check the previous article), checked the statistics, and saw that the Cache full count was growing. My cache had filled several times.

According to the APC configuration documentation there are two settings that control the expiration of cache entries. They are:

Using PHP APC with Cherokee

Sun, 28 Nov 2010 00:00:00 +1100

If you’re using Cherokee as your web server, you’re probably already interested in squeezing every drop of performance out of your server. If you’re hosting PHP sites with Cherokee, using APC (Alternative PHP Cache) could enable you to squeeze out even more drops.

The Alternative PHP Cache (APC) is a free and open opcode cache for PHP. Its goal is to provide a free, open, and robust framework for caching and optimizing PHP intermediate code.

Using ISPConfig with Cherokee

Thu, 25 Nov 2010 00:00:00 +1100

The Cherokee Project describes Cherokee as:

a very fast, flexible and easy to configure Web Server. It supports the widespread technologies nowadays: FastCGI, SCGI, PHP, CGI, uWSGI, SSI, TLS and SSL encrypted connections, Virtual hosts, Authentication, on the fly encoding, Load Balancing, Apache compatible log files, Data Base Balancing, Reverse HTTP Proxy, Traffic Shaper, Video Streaming and much more.

It makes a great alternative to the swiss-army chainsaw that is Apache and the documentation, wizards, and screencasts make it infinitely more friendly than nginx.

My Djangy invite arrived!

Tue, 23 Nov 2010 00:00:00 +1100

My Djangy invite arrived and I’m very excited!

Now I just need to set aside the time to use it. I already have some ideas…

Install uWSGI (latest) on Ubuntu Server 10.10

Sun, 21 Nov 2010 00:00:00 +1100

uWSGI is a fast, self-healing, WSGI server. It is typically used with Python web applications. It works very well with the Cherokee Web Server and the Django web application framework.

To install the latest uWSGI use pip and the URL for the latest version of uWSGI:

```bash
sudo apt-get install libxml2-dev build-essential python-dev python-pip
sudo pip install http://projects.unbit.it/downloads/uwsgi-latest.tar.gz



Simple as that.

You should [follow me on Twitter](https://twitter.com/davidltaylor).

Adding the Cherokee Web Server PPA to Ubuntu

Sat, 20 Nov 2010 00:00:00 +1100

The following commands will add the Cherokee Web Server PPA to Ubuntu (version 9.10 and higher). The first command adds the extremely handy add-apt-repository program to your system. The second adds the Cherokee Personal Package Archive.

```bash
sudo apt-get install python-software-properties
sudo add-apt-repository ppa:cherokee-webserver/ppa



At the moment, Cherokee is at version 1.0.9.

If you don't already have Cherokee installed and you want to install it I find the following combination of packages most useful:


 ```bash
 sudo apt-get update
 sudo apt-get install cherokee cherokee-doc libcherokee-mod-libssl libcherokee-mod-streaming libcherokee-mod-rrd libcherokee-mod-admin spawn-fcgi

That’s it. If you want more simple posts on building lean servers, cloud computing, software development and more, subscribe to my RSS feed and/or follow me on Twitter. Cheers!

How to upgrade Ubuntu Server 10.04 to 10.10

Fri, 19 Nov 2010 00:00:00 +1100

The following commands can be followed blindly to update an Ubuntu Server install from version 10.04 (Lucid) to version 10.10 (Maverick).

```bash
apt-get update
apt-get install update-manager-core
sed -i 's/Prompt=lts/Prompt=normal/' /etc/update-manager/release-upgrades
do-release-upgrade

Google App Engine SDK broken on Ubuntu Maverick 10.10

Sun, 24 Oct 2010 00:00:00 +1100

The Google App Engine SDK is broken on Ubuntu 10.10, but it’s easy enough to fix…

The Google App Engine SDK requires Python 2.5. It does not work with Python 2.6.

This problem is covered in these Google Code issues:

If you try to use Google App Engine SDK on Python 2.6 you will encounter tracebacks like the following:

```text
Traceback (most recent call last):
 File "/home/david/lib/google_appengine/google/appengine/tools/dev_appserver.py", line 3211, in _HandleRequest
 self._Dispatch(dispatcher, self.rfile, outfile, env_dict)
 File "/home/david/lib/google_appengine/google/appengine/tools/dev_appserver.py", line 3154, in _Dispatch
 base_env_dict=env_dict)
 File "/home/david/lib/google_appengine/google/appengine/tools/dev_appserver.py", line 527, in Dispatch
 base_env_dict=base_env_dict)
 File "/home/david/lib/google_appengine/google/appengine/tools/dev_appserver.py", line 2452, in Dispatch
 CGIDispatcher.Dispatch(self, *args, **kwargs)
 File "/home/david/lib/google_appengine/google/appengine/tools/dev_appserver.py", line 2404, in Dispatch
 self._module_dict)
 File "/home/david/lib/google_appengine/google/appengine/tools/dev_appserver.py", line 2441, in curried_exec_cgi
 return ExecuteCGI(*args, **kwargs)
 File "/home/david/lib/google_appengine/google/appengine/tools/dev_appserver.py", line 2312, in ExecuteCGI
 logging.debug('Executing CGI with env: %s', pprint.pformat(env))
 File "/usr/lib/python2.6/pprint.py", line 60, in pformat
 return PrettyPrinter(indent=indent, width=width, depth=depth).pformat(object)
 File "/usr/lib/python2.6/pprint.py", line 119, in pformat
 self._format(object, sio, 0, 0, {}, 0)
 File "/usr/lib/python2.6/pprint.py", line 137, in _format
 rep = self._repr(object, context, level - 1)
 File "/usr/lib/python2.6/pprint.py", line 230, in _repr
 self._depth, level)
 File "/usr/lib/python2.6/pprint.py", line 242, in format
 return _safe_repr(object, context, maxlevels, level)
 File "/usr/lib/python2.6/pprint.py", line 284, in _safe_repr
 for k, v in _sorted(object.items()):
 File "/usr/lib/python2.6/pprint.py", line 75, in _sorted
 with warnings.catch_warnings():
 File "/usr/lib/python2.6/warnings.py", line 333, in __init__
 self._module = sys.modules['warnings'] if module is None else module
 KeyError: 'warnings'


... which is a pain in the proverbial to debug.

Ubuntu Maverick 10.10 (and some earlier releases of Ubuntu) uses Python 2.6 by default. In older versions of Ubuntu the fix involved installing the older Python 2.5 from the official repository. However, with Ubuntu Maverick 10.10 things are worse... Python 2.5 is no longer available from an official repository!

Thankfully, there is a repository on Launchpad that contains the necessary packages. To install, do the following:


 ```bash
 sudo add-apt-repository ppa:fkrull/deadsnakes
 sudo apt-get update
 sudo apt-get install python2.5

Next, modify dev_appserver.py in the Google App Engine SDK. Change the first line from:

RightScale ServerTemplate: Bitnami Drupal Stack

Thu, 21 Oct 2010 00:00:00 +1100

I’ve published a RightScale ServerTemplate for installing Bitnami’s Drupal stack on Amazon EC2.

Drupal is an open source content management platform powering millions of websites and applications. It’s built, used, and supported by an active and diverse community of people around the world.

My ServerTemplate is based on an existing ServerTemplate by Bitnami. It differs from Bitnami’s existing Drupal ServerTemplate in a few ways:

it uses an Ubuntu RightScale MultiCloud Image as its base
it has some of RightScale’s RightScripts for logging and monitoring added
it installs the Drupal stack from a supplied URL, meaning you can trial different versions of the Drupal stack easily

Using this ServerTemplate you can easily fire up a Drupal instance to either:

RightScale ServerTemplate: Bitnami WordPress Stack

Thu, 21 Oct 2010 00:00:00 +1100

I’ve published a RightScale ServerTemplate for installing Bitnami’s WordPress stack on Amazon EC2.

WordPress is web software you can use to create a beautiful website or blog.

It is based on an existing ServerTemplate by Bitnami. It differs from Bitnami’s existing WordPress ServerTemplate in a few ways:

it uses an Ubuntu RightScale MultiCloud Image as its base
it has some of RightScale’s RightScripts for logging and monitoring added
it installs the WordPress stack from a supplied URL, meaning you can trial different versions of the WordPress stack easily

Using this ServerTemplate you can easily fire up a WordPress instance to:

RightScale ServerTemplate: Virtualmin

Wed, 20 Oct 2010 00:00:00 +1100

I’ve published a RightScale ServerTemplate for launching Virtualmin on Amazon EC2.

Virtualmin is a powerful and flexible web server control panel based on the well-known Open Source web-based systems management GUI, Webmin. Manage your virtual domains, mailboxes, databases, applications, and the entire server, from one comprehensive and friendly interface.

The ServerTemplate I’ve created makes it easy for anyone to get started providing managed hosting on cloud infrastructure.

Migrating from Google App Engine to Django

Mon, 18 Oct 2010 00:00:00 +1100

Unfortunately, I had to migrate one of my projects from Google App Engine to Django on a self-managed server. I didn’t want to do this. See Google App Engine Gotcha #2 and Google App Engine Gotcha #3 for the main reasons. Those headaches became bad enough that I had to bite the bullet and migrate out of Google App Engine.

It’s not for the faint-hearted…

Why Google App Engine In The First Place? #

Google App Engine is a PaaS (Platform as a Service). PaaS products eliminate the need for lower-level server management and, if they’re good, make it very easy to write and maintain code for that platform.

Hudson: pam_authenticate failed

Mon, 11 Oct 2010 00:00:00 +1100

If you’re getting login failures after enabling Unix user/group database security check the log: /var/log/hudson/hudson.log.

If you see:

```text
Oct 11, 2010 9:41:08 PM hudson.security.AuthenticationProcessingFilter2 onUnsuccessfulAuthentication
INFO: Login attempt failed
org.acegisecurity.BadCredentialsException: pam_authenticate failed : Authentication failure; nested exception is org.jvnet.libpam.PAMException: pam_authenticate failed : Authentication failure
 at hudson.security.PAMSecurityRealm$PAMAuthenticationProvider.authenticate(PAMSecurityRealm.java:100)
 at org.acegisecurity.providers.ProviderManager.doAuthentication(ProviderManager.java:195)
 at org.acegisecurity.AbstractAuthenticationManager.authenticate(AbstractAuthenticationManager.java:45)
 at org.acegisecurity.ui.webapp.AuthenticationProcessingFilter.attemptAuthentication(AuthenticationProcessingFilter.java:71)
 at org.acegisecurity.ui.AbstractProcessingFilter.doFilter(AbstractProcessingFilter.java:252)
 at hudson.security.ChainedServletFilter$1.doFilter(ChainedServletFilter.java:87)
 at org.acegisecurity.ui.basicauth.BasicProcessingFilter.doFilter(BasicProcessingFilter.java:173)
 at hudson.security.ChainedServletFilter$1.doFilter(ChainedServletFilter.java:87)
 at org.acegisecurity.context.HttpSessionContextIntegrationFilter.doFilter(HttpSessionContextIntegrationFilter.java:249)
 at hudson.security.HttpSessionContextIntegrationFilter2.doFilter(HttpSessionContextIntegrationFilter2.java:66)
 at hudson.security.ChainedServletFilter$1.doFilter(ChainedServletFilter.java:87)
 at hudson.security.ChainedServletFilter.doFilter(ChainedServletFilter.java:76)
 at hudson.security.HudsonFilter.doFilter(HudsonFilter.java:164)
 at winstone.FilterConfiguration.execute(FilterConfiguration.java:195)
 at winstone.RequestDispatcher.doFilter(RequestDispatcher.java:368)
 at winstone.RequestDispatcher.forward(RequestDispatcher.java:333)
 at winstone.RequestHandlerThread.processRequest(RequestHandlerThread.java:244)
 at winstone.RequestHandlerThread.run(RequestHandlerThread.java:150)
 at java.lang.Thread.run(Thread.java:619) Caused by: org.jvnet.libpam.PAMException: pam_authenticate failed : Authentication failure
 at org.jvnet.libpam.PAM.check(PAM.java:105)
 at org.jvnet.libpam.PAM.authenticate(PAM.java:123)
 at hudson.security.PAMSecurityRealm$PAMAuthenticationProvider.authenticate(PAMSecurityRealm.java:90)
 ... 18 more

… the problem is likely caused by Hudson being unable to read your /etc/shadow file.

Cherokee failing to exec spawn-fcgi

Mon, 04 Oct 2010 00:00:00 +1100

If you’re using spawn-fcgi with Cherokee (probably to get a wiki or PHP application or similar working) and you see:

503 Service Unavailable

… you might be having the same problem I had. That is, Cherokee was failing to launch spawn-fcgi.

To verify this is the cause of your problem, stop Cherokee, then start it manually at the command line:

# /usr/sbin/cherokee --admin_child -C /etc/cherokee/cherokee.conf
Cherokee Web Server 1.0.8 (Aug 18 2010): Listening on port ALL:80, TLS disabled, IPv6 enabled, using epoll, 4096 fds system limit, max. 2041 connections, caching I/O, 20 threads, 102 connections per thread, standard scheduling policy

Next, try to visit the site giving you the problem. If you see the following in your terminal then you’ve probably got the same problem I had:

Install Hudson on Debian Lenny for Continuous Integration

Mon, 04 Oct 2010 00:00:00 +1100

I found that Hudson depends on daemon but it doesn’t install automatically. So, before we get started:

```bash
apt-get install daemon



Now that's out of the way, we need to grab the Hudson key and install the package:


 ```bash
 wget -q -O - http://hudson-ci.org/debian/hudson-ci.org.key | apt-key add -
 cd /tmp
 wget http://hudson-ci.org/latest/debian/hudson.deb
 dpkg --install ./hudson.deb

If everything went well you’ll see Hudson running:

ps auxw | grep hudson
hudson 9101 0.0 0.1 2108 516 ? Ss 09:59 0:00 /usr/bin/daemon --name=hudson --inherit --env=HUDSON_HOME=/var/lib/hudson --output=/var/log/hudson/hudson.log --pidfile=/var/run/hudson/hudson.pid -- /usr/bin/java -jar /usr/share/hudson/hudson.war --webroot=/var/run/hudson/war --httpPort=8080 --ajp13Port=-1
hudson 9103 18.9 10.3 297064 52660 ? Sl 09:59 0:03 /usr/bin/java -jar /usr/share/hudson/hudson.war --webroot=/var/run/hudson/war --httpPort=8080 --ajp13Port=-1

… and listening on port 8080:

Install uWSGI (latest) on Debian Lenny

Mon, 04 Oct 2010 00:00:00 +1100

First, why would you want it? uWSGI is a fast, self-healing, WSGI server, originally intended for use with Python web applications. I intend to use it with Python Django, served by Cherokee.

As for installation…

Do not bother doing a hands-on install from source. It’s messy. Just use pip:

```bash
apt-get install gcc python-dev libxml2-dev
apt-get install python-pip
pip install http://projects.unbit.it/downloads/uwsgi-latest.tar.gz



Simple as that.

You should [follow me on twitter](https://twitter.com/davidltaylor).

Python One-Liner: Debug Mail Server

Sun, 03 Oct 2010 00:00:00 +1000

Need a pretend mail server that you can use for debugging? Try this:

```bash
python -m smtpd -n -c DebuggingServer localhost:1025



It listens on port 1025 for local connections.

Install Cherokee (testing) on Debian Lenny

Sat, 02 Oct 2010 00:00:00 +1000

Add the following to /etc/sources:

```text
deb http://ftp.us.debian.org/debian/ testing main contrib non-free



Create `/etc/apt/apt.conf` with:


 ```text
 APT::Default-Release "stable";

Create /etc/apt/preferences for pinning the testing packages:

```text
Package: libssl-dev
Pin: release a=testing
Pin-Priority: 999 
Package: libssl0.9.8
Pin: release a=testing
Pin-Priority: 999 
Package: libcherokee-base0
Pin: release a=testing
Pin-Priority: 999 
Package: mysql-common
Pin: release a=testing
Pin-Priority: 999 
Package: libmysqlclient16
Pin: release a=testing
Pin-Priority: 999 
Package: libcherokee-server0
Pin: release a=testing
Pin-Priority: 999 
Package: libcherokee-config0
Pin: release a=testing
Pin-Priority: 999 
Package: libcherokee-mod-admin
Pin: release a=testing
Pin-Priority: 999 
Package: cherokee
Pin: release a=testing
Pin-Priority: 999 
Package: *
Pin: release a=stable
Pin-Priority: 500



Then install the packages:


 ```bash
 apt-get install -t testing cherokee libcherokee-base0 libcherokee-server0 libcherokee-config0 libcherokee-mod-admin libssl0.9.8

That’s the installation. For the configuration, check out the Cherokee cookbooks.

Migrating from Google App Engine to Django... a preamble

Wed, 29 Sep 2010 00:00:00 +1000

I’ve been nose-deep migrating a project from Google App Engine to Django. See Google App Engine Gotcha #2 and Google App Engine Gotcha #3 for the reasons. I’m keeping notes on the experience, in case any other poor suckers out there have to migrate from the cosy, loving embrace of a PaaS to the cold, hard trenches of a self-managed server.

Stay tuned…

IronRuby on Rails on Azure (Part 2 of 3) – Connecting your Rails app to SQLAzure

Wed, 08 Sep 2010 00:00:00 +1000

This is the continuation of my bungling around with IronRuby, Rails and getting it running on SLQAzure. In part 1, we installed a basic rails application using IronRuby with SQL Server as it’s backend. We’ll now examine how we can connect this application to a SQLAzure backend as opposed to a SQL Server hosted on your development machine.

I’m assuming you have the vanilla rails application from Part 1 up and running and have verified that the rails application can communicate with your SQL server database by clicking on the “About your application’s environment” link as described in the previous article. I’m also assuming you’ve already signed up for a SQLAzure account and know how to connect to it.

IronRuby on Rails on Azure (Part 1 of 3) - Getting Started

Tue, 07 Sep 2010 00:00:00 +1000

This is the first of a three part series on running a rails application on Windows Azure using SQLAzure as your database backend. The Rail Guides website, of course, has a wonderful getting started article that I couldn’t possibly do justice. This small series is just a very simple demonstration of one possible way of developing a rails application on Microsoft’s Azure platform.

In this part, we’ll simply be focusing on getting a basic rails application running on your laptop or desktop using IronRuby and SQL Server and links to download bundles that you can try out for yourself. If you already know how to do all that, skip to the end of this article to the download bundles and head over to Part 2: Connecting your rails application to SQLAzure

Google App Engine Gotcha 3

Sun, 29 Aug 2010 00:00:00 +1000

What’s the problem? #

Well, gotcha #3 is a simple but annoying one. The urlfetch call does not use a fixed source IP address.

It makes sense, when you think about it. Google App Engine is (apparently) distributed all over the place, with no guarantee of where your code will run at any given time. I presume this is so that Google are free to move applications between data centres to optimise uptime and performance.

Google App Engine - unknown URL handler type

Sun, 15 Aug 2010 00:00:00 +1000

 Fatal error when loading application configuration:
 
 Invalid object:
 Unknown url handler type.
 <URLMap 
 secure=always 
 static_files=None 
 auth_fail_action=redirect 
 require_matching_file=None 
 static_dir=None 
 script=None 
 url=https://cloudartisan.com/admin/.* 
 upload=None 
 expiration=None 
 login=admin 
 mime_type=None
 >
 in "/home/david/Subversion/cloudzu/software/cloudzuum/src/app.yaml", line 37, column 1

 
This happens when you have an error in your `script:` definition in `app.yaml` or `script:` is missing entirely.

In my case, it was missing entirely... woops!

appengine-jruby "no such file to load -- builtin/core_ext/symbol

Fri, 13 Aug 2010 00:00:00 +1000

A while ago I wrote a proof of concept in JRuby on Google App Engine, using the fantastic appengine-jruby project. The code worked (albeit, very slowly!).

Satisfied, I put the code to the side, committed to a Subversion repository.

Today I got a chance to dust off the code and do some more work on it… only to find it no longer worked.

When I started my code with dev_appserver.rb I saw:

Google App Engine Gotcha 2

Mon, 09 Aug 2010 00:00:00 +1000

For Google App Engine: Gotcha #2 I choose the default 5 second timeout on urlfetch. This function is part of the Google App Engine’s API and applies to both Python and Java.

The Python documentation states:

The deadline can be up to a maximum of 10 seconds. If deadline is None, the deadline is set to 5 seconds.

The Java documentation states:

You can set a deadline for a request, the most amount of time the service will wait for a response. By default, the deadline for a fetch is 5 seconds. The maximum deadline is 10 seconds.

Auto-Scaling Is Avoidable, Potentially Dangerous... But Fun

Sat, 07 Aug 2010 00:00:00 +1000

RightScale have some tutorials on using their service to set up auto-scaling. Recently I played around with it, configuring auto-scaling in response to increases and decreases in server load. Honestly, it was fun. But it was essentially a folly, an exercise in dealing unnecessarily with dangerous corner cases.

Here’s why…

Unconstrained Scaling #

Unconstrained scaling is bad, mmkay. If you allow your systems to scale without constraint, you’re risking a surprise credit card bill. Worse yet, if you’re not paying any attention to your systems (shame!) then you could be in for a jaw-dropping credit card bill.

CloudCamp Sydney 2010v2 (Aug 6, 2010)

Fri, 06 Aug 2010 00:00:00 +1000

I’ve just come back from CloudCamp Sydney 2010v2 and I have some mixed thoughts about the event…

Event Format #

The event format is an unconference.

An unconference is a facilitated, participant-driven conference centered on a theme or purpose.

Basically, the format is an effort to break away from the staid conferences of old, where the audience would fight boredom and stave off sleep by hyper-caffeinating themselves and perching uncomfortably on plastic seats. For the most part, the format worked very well.

Auditing Cloud Service Providers

Tue, 27 Jul 2010 00:00:00 +1000

What’s Up? #

I want to find existing documented best practice processes and questionnaires for auditing cloud service providers.

I was hoping to find existing surveys/questionnaires that can be reused to perform audits on cloud service providers. I was also hoping those audits could be performed without introducing much additional load for the auditor or the cloud service providers. Especially for the auditor… because, no doubt, that will be me.

Google App Engine Gotcha 1

Sun, 27 Jun 2010 00:00:00 +1000

I like Google App Engine. However, there are a number of gotchas that can creep up on you. Especially if you dive right in without doing any reading first.

Here’s gotcha #1…

A number of standard Python modules are not available or only provide limited functionality. Click here to see Google App Engine’s list of enabled, partially-enabled and empty modules.

Sometimes developing for Google App Engine isn’t straightforward and requires a little imagination. However, most of the time there’s a simple alternative; it just takes a little reading. For example, if you want to send e-mail on Google App Engine you can’t use Python’s smtplib (because it relies on socket, which is implemented as an empty module). Instead, you would do the following:

Install EC2 AMI & API Tools in Debian

Wed, 02 Jun 2010 00:00:00 +1000

Download the zip files for the tools:

```bash
$ wget http://s3.amazonaws.com/ec2-downloads/ec2-api-tools.zip
$ wget http://s3.amazonaws.com/ec2-downloads/ec2-ami-tools.zip


Unzip the tools:

 ```bash
 $ unzip ec2-api-tools.zip
 $ unzip ec2-ami-tools.zip

This will give you some directories with version numbers appended. Put them in a safe place:

```bash
$ mkdir /usr/local/ec2
$ cp -r ec2-ami-tools-1.3-34544/* /usr/local/ec2/
$ cp -r ec2-api-tools-1.3-42584/* /usr/local/ec2/


Install Java:

 ```bash
 $ apt-get install sun-java6-jre

Edit the system-wide profile:

```bash
$ vi /etc/profile


Add the following to the file:

 ```bash
 # EC2 Tools
 export EC2_HOME=/usr/local/ec2
 export PATH=$PATH:$EC2_HOME/bin
 export JAVA_HOME=/usr

When someone logs in, their environment will be ready and raring to go. Users can then issue commands such as:

Scheduled Tasks With Google App Engine & Python

Wed, 02 Jun 2010 00:00:00 +1000

We’ll create a simple Hello World application that sends an e-mail every 5 minutes to reassure you that the Internet is still out there and still cares.

First, download and install the Google App Engine SDK.

For example:

```bash
wget http://googleappengine.googlecode.com/files/google_appengine_1.3.4.zip
unzip google_appengine_1.3.4.zip



Create a directory for your application:

 ```bash
 mkdir helloworld

Create an app.yaml file to describe your application:

```yaml
application: helloworld
version: 1
runtime: python
api_version: 1handlers:
- url: /helloworld
 script: helloworld.py
 login: admin


Create a cron.yaml file to run your scheduled task:

 ```yaml
 cron:
 - description: hello... is it me you're looking for?
 url: /helloworld
 schedule: every 5 minutes

Create your script (changing the e-mail addresses, of course):

RightScripts In Any Scripting Language

Sun, 30 May 2010 00:00:00 +1000

RightScale’s cloud management platform can be used to deploy and manage solutions on cloud infrastructure with some degree of automation, control, and scaling.

Users create RightScripts to automate actions on their servers. Think of them the same way you would any other script for managing a server. The differences are that they are managed inside RightScale’s management platform, with version control, easy reuse, and access to handy environment variables. A RightScript is typically written in Bash, Ruby, or Perl, and has variables that are initialised using input parameters.