Category Archives: Open Source

Docker: Second Birthday

Docker is one of the hottest technologies; I started with it’s first release not too long back, however, in Docker world it feels like Docker has been around for a long time.

What a paradigm shift we have witnessed: a few technologies get that kind of attention, momentum, growth, and community in such a short time (24 months).

Today, we use Docker to develop, test, ship, integrate, deploy, and run apps/services. Docker has solved developer (devops) usability problems; it’s hard to go back and do things without Docker.

Haven’t heard of Docker, or didn’t find time to try it out? Start with following links.

What’s new with Docker? Look at following links:

  • Docker Compose (docker-compose, formerly known as fig)

    Compose is a tool for defining and running complex applications with Docker. With Compose, you define a multi-container application in a single file, then spin your application up in a single command which does everything that needs to be done to get it running.

  • Docker Machine (docker-machine)

    Machine makes it really easy to create Docker hosts on your computer, on cloud providers and inside your own datacenter. It creates servers, installs Docker on them, then configures the Docker client to talk to them.

  • Docker Swarm (swarm)

    Docker Swarm is native clustering for Docker. It turns a pool of Docker hosts into a single, virtual host.Swarm serves the standard Docker API, so any tool which already communicates with a Docker daemon can use Swarm to transparently scale to multiple hosts: Dokku, Compose, Krane, Flynn, Deis, DockerUI, Shipyard, Drone, Jenkins… and, of course, the Docker client itself.

Docker: Automatically Restarting Containers

As a Docker beginner, I always wondered, how to automatically restart container os or system boot?

Thankfully, Docker 1.2 introduced restart policies (which makes sure even linked containers are started in expected order).  The --restart=always flag/value can be passed to ensure automatic restart of containers.

docker run --restart=always redis

Read more about Docker Restart Policies, or other methods.

Github: Forking and Contributing Workflow

I stick to following workflow while contributing to open-source projects on Github:

  1. Fork <user>/<repo> to abdul/<repo>
  2. git clone [email protected]:abdul/<repo>.git
  3. git remote add upstream [email protected]:<user>/<repo>.git
  4. git checkout -b <topic_branch>
  5. Fix, Add or Change things (code, documentation, etc.)
  6. git add <files>
  7. git commit -m <commit-message>
  8. Repeat #5 , #6 & #7
  9. git pull --rebase upstream master
  10. Resolve any conflicts, if required
  11. Squash Commits, e.g. squashing last three commits:
    • git rebase -i HEAD~3
  12. git push origin <topic_branch>
  13. Go to<repo>
  14. Create a Pull Request [Github is smart enough to detect a change in forked repo, and offer you to compare and create pull-request]
  15. Discuss changes in PR, and Fix things in branch as required.


NodeSchool is a fully community driven effort and is 100% open source. This means that anyone can use the NodeSchool materials and/or host an event.

I have set up NodeSchool Kanpur chapter, and I hope to organise (with help of like minded people) a couple of workshops on Node.js (and web-development ecosystem around it).

I am looking for mentors and co-organisers, if you are interested, let’s have a conversion.


Phabricator (pronounced like the word fabricator) is a suite of web applications which make it easier to build software, particularly when working with teams. Phabricator is largely based on Facebook’s internal tools.

It has been more than three years since I have been using Phabricator (self-hosted) at work. It is one of the open-source projects which has inspired me a lot. I have seen it grow over these years; every day it gets better. Thanks to hundreds of open-source developers for their contribution.

Want to see how to develop elegant and sophisticated software, even with a poor programming languages like PHP? Look at Phabricator’s source-code and tools (built using PHP), you would be awestruck.

Want to play with Phabricator without installing? Have a look at PhorePlay (Phabricator hosting service), which offers free Phabricator hosting (with multiple instances) up to five users.  I think, that’s good enough for most of us (small teams).

Phabricator’s core team is working to launch Phacility (Phabricator Hosting). I think, it has potential to become something like Github (useful and essential service) one day.

Installing NumPy and SciPy on Mac OSX 10.8 (Mountain Lion)

I am playing with ExpEyes, which is awesome tool for anyone who wants to learn electronics (and physics).

ExpEyes comes with software (written in python) for GNU/Linux and Windows. I intend to run it on Mac OSX, because it can (provided all dependencies are met).

It is generally easy to install python packages on OSX using easy_install or pip, however, installing NumPy and SciPy turned out to be pain on OSX (10.8.2, latest version the time of posting).

In my case, I am using python binary installed via homebrew, so I can’t take advantage of numpy (which comes installed with mountain-lion).

I have XCode 6 installed, which comes with gcc-4.2.1 (llvm build). I figured out, I can build numpy and scipy from source.

Following is what I did to build numpy and scipy. SciPy requires gfortran, so that has to be installed (I used homebrew to do that):

$ git clone
$ git clone
$ cd numpy
$ python build && python install
$ brew install gfortran
$ cd ../scipy
$ python build && python install

BTW! In case, you are wondering, why didn’t I use Scipy Superpack script, which indeed makes job easier?

I didn’t, because I want to install gnu/unix packages using homebrew, and I already have git and other things. Super Spicypack downloads (gfortran, etc.), builds and installs packages directly to system, making it harder to remove those later? Homebrew makes job easier.

IndieReign – Discover a world of independent film and have fun doing it

My team and I have been working on bunch of products/services for last 3-4 years.

IndieReign is the one, we have been mostly focused on, and took care entire engineering and technical architecture.

It is a platform for indie filmmakers to promote, distribute and monetize content. It is actually more than that.

IndieReign runs on open-source software stack (for everything including video transcoding) with exception of Adobe Flash Player (used for video playback), and Amazon’s cloud.

Please share your feedback.

Mac OSX (10.8) Mountain Lion: Fixing Apache, PHP, Subversion, User Sites, Virtual Hosts Directories

While Mountain Lion (OSX 10.8) has brought many enhancements and fixes (mostly performance), it has also caused frustration to old users.

You might have noticed following issues in Mountain Lion:

  1. Web Sharing Preferences Panel has been removed, i.e. you need to launch apache manually or write some launch-agent script or hack to bring web-sharing preferences panel back
  2. User Sites (per-user web directories) is disabled, i.e. http://localhost/~username doesn’t work by default, and you need to modify apache config file to enable it
  3. PHP extension for Apache is disabled by default i.e. you need to enable it by modified apache config (httpd.conf)
  4. Subversion is not installed by default, either you can install using homebrew or by install XCode command line tools

You can fix these issues (1-4) by following these two links:

Whenever I upgrade to newer version of OSX, I end up with broken development-setup i.e. custom permissions on directories are messed up, configuration files are overwritten/removed, etc.

I think, I have found a way to keep default directory permissions, so future upgrades don’t break things.

My development setup requires my virtual hosts to use directories within my home-directory (and outside of ~/Sites or /Library/WebServer/Documents directories).

I am using following technique to setup my virtual-hosts:

  • Enable User Sites  (i.e. fix #2 problem above)
  • Create symbolic-links(symlinks) of your project folder(s) inside ~/Sites directory
  • Enable FollowSymLinks and SymLinksIfOwnerMatch options in virtual-host configuration.

In my case, I have a project inside ~/projects/myproject . I create a symlink of this directory inside ~/Sites, like this using Terminal (command-prompt)

$ ln -s ~/projects/myproject ~/Sites/myproject

I have following virtual-host configuration for this project:

<VirtualHost *:80>
 DocumentRoot "/Users/myusername/Sites/myproject/trunk/web"
 ServerName  myproject
 ServerAdmin myusername@localhost
 Options Indexes MultiViews Includes +FollowSymlinks +SymLinksIfOwnerMatch
 CustomLog "/private/var/log/apache2/myproject_access.log" common
 ErrorLog "/private/var/log/apache2/myproject_error.log"

I use it on my development machine (i.e. my laptop), so I don’t really bother about security at Apache level, and I take care of security at firewall/other levels.

If you are going to use above technique/configuration, please understand everything (by reading documentation) before you use it.

This set up works for me and solves some problems for me – I don’t have to change default directory permissions, and I don’t have to keep my projects within ~/Sites or /Library/WebServer/Documents directory.

RIP, Kenneth Gonsalves “KG” (1953 — 2012)

Kenneth Gonsalves(Image Courtesy: Linux Bangalore (

Kenneth Gonsalves, a champion in Indian (NRC)FOSS, Python and Django community passed away on August 3, 2012. Kenneth’s presence would be hugely missed in the community.

In 2008, Tarun Dua introduced me to Kenneth. I interacted with him at various occasions about NRCFOSS and general stuff. I was very inspired to see his contributions, specially some of his open-source projects.

While checking out NRCFOSS website today, I came to know about this sad news.

May his soul rest in peace.

Installing Riak on Mac OSX

I am evaluating various nosql solutions for our existing/new projects. Bluesmoon recommended about Riak which is an awesome nosql datastore/database (and lot more).  I couldn’t wait to check it out.

As usual, I relied on Homebrew – the best package manage for OSX – to install applications/libraries. While installing, I got an error:

$ brew install riak
$ ...
$ Error: Failed executing: make all rel

After searching for a while, I figured out that there are known issues. With the help of Adam‘s comment on github, I was able to install Riak finally.

We make brew ignore Riak’s dependencies while building it.

Following commands should build and install Riak:

$ brew update
$ brew install erlang
$ brew install --HEAD --ignore-dependencies riak -v

Thoughts: Open Standards and Technology for Movie Rental Business

I have been thinking about open-standards for movie-rental business. I searched to find about such standards but I couldn’t locate. Do you know of any?

Basically, a set of specifications/protocols which can be implemented by different players in movie-rental chain. Players include studios, distributor, application-developers, portals, et. al.

Rough thoughts:

  • A studio/distributor implements this specification/protocol and exposes API  for metadata/streaming(or playback)/file-access from their own server
  • Third party developers use the API to to pull metadata and play the content
  • Third party developers can charge the user and split the revenue with content owner
  • Or, studios/distributors can enable payment-gatway APIs as part of their offering
  • Content is always played directly from studio/distributor’s server
  • Studio/distributor have total control over their assets and also charge based on subscription/per-access
This would solve many problems (billing, stats, etc.), however, there would be concerns:
  • someone (end-user, third-party developer) might try to cache/save streams as file and distribute
  • and other concerns?
Once we have specifications and protocols in place – a stack using open-source technologies can be built as template, and made available so that everyone in chain can use/customize it to build their part.
I wanted to talk about it on twitter but 140 chars were too short to express my thoughts. Anyway, it’s good to have a blog post in place.
Please share your views.

Development Setup – Updated

I posted about our proposed development setup, during these years a few things have changed a bit. More or less, this is what we have:

  • GNU/Linux (Ubuntu) and Windows(some) boxes for developers
  • Eclipse (PDT) IDE with loads some plugins
  • Version Control: Subversion, GIT, GIT-SVN
  • Browsers
    • Firefox
    • Chrome
    • Opera
    • Safari
    • OperaMini (Simulator)
    • FirefoxMobile (Simulator)
  • Plugins for browsers:
    • Firebug (with loads of plugins – YSlow, etc.)
    • Web Developers Tools (firefox)
    • Others (developer’s preference)
  • Developers also use WINE to run some windows software on GNU/Linux


  • Windows Box with different browsers (versions, vendors)
  • Windows Box with different Mobile Emulators/Simulators
  • Selenium Plugin for Firefox

Development Servers/Applications:

  • Apache HTTPD
  • Mysql Server
  • Trac
  • MediaWiki
  • PHPMyAdmin
  • ReviewBoard (limited usage)
  • Hudson (limited usage)

Virtualization with VMware vSphere (ESXi) Hypervisor

One of our servers, collocated in our office, crashed couple of weeks back. We took this opportunity to virtualize most of the stuff, earlier only some intances were virtualized.

Installing and setting up VMware vSphere Hypervisor is a straight forward process, even the process of creating virtual-machines.

Most of time, I spent was figuring out:

  • Storage devices and configuration
  • How many virtual machines, do we need?
  • Which operating systems, do we need?
  • Network configuration

I am not going to cover details of installation and network configuration. I might write more posts in future to cover specifics. Meanwhile, I would love to answer the comments, if I know or share anything I know.

Anyway, this is what we have now:

  • IBM x3250 M2
    • Intel Xeon (Quadcore) x64
    • 4 GB memory
    • 3TB+ storage connected to LSI RAID and SATA controller
  • Host: VMware vSphere Hypervisor – which runs out of  1.5 TB datastore, directly connected to SATA controller (different from LSI)
  • Guest VMs:
    • OpenVPN Access Server for VPN
    • Ubuntu 10.10 Server as Development sandbox
    • Zentyal(eBox) as Internet Gateway (filtering, IDs, proxy)
    • Ubuntu 10.04 LTS Server for SVN, Mails, Local Staging WebServer
    • Windows 2008 Server for ActiveDirectory, FileServer, Windows-Specific-Software (access-control-management-gui, etc.)
    • Others: Project Specific VMs

We are very happy with the improved performance and stability, and the way hardware is properly used. Thanks to VM templates, clones and snapshots, we would be able to provision above infrastructure with minimal downtime.

I would soon replace Windows Server with Zentyal (eBox), which I believe, can handle most of things. However, we would need another Windows XP box to run some software (proprietary), which only runs on windows, to manage various other devices (EPBX, Access-Control-System, etc.).

We now have relatively more scalable infrastructure, more virtual machines can be created to take care of email-server, svn-server, etc.

We are yet to invest in a sophisticated storage device such as DAS or NAS or SAN, hence we are relying on some cheap tricks here:

  • 1.5 TB disk as datastore for VMware vSphere, which stores
    • Guest Operating System VMs
    • Guest Operating System VMs Paging files
    • Virtual Disks (non critical data – programs and configs) required by host virtual-machines
  • 1.5 TB disk via Raw Device Mapping (RDM) shared among virtual machines
  • 500 GB disk via Raw Device Mapping (RDM) shared among virtual machines
  • 1.5 TB USB/Firewire – Backup storage
    • Rsync is used to backup critical data and configs from all virtual machines on daily, weekly and monthly basis.

Things I didn’t like:

  • VMware vSphere client requires Microsoft Windows
  • VMware vSphere client doesn’t allow creating RDMs from GUI, one has to use these instructions

Finally, I could execute some of things I planned more than two years back.

CalDAV and SyncML Resources

One of our projects requires a calendar application with bi-directional synchronization with any other calendar applications/services such as Microsoft Outlook, Mozilla Thunderbird, Google Calendar, Yahoo! Calendar, iPhone, Nokia or others.

We can export iCalendar file, that would be readonly i.e. client applications can not write back to source calendar

I started looking at CalDAV, which is an extension to WebDAV, and SyncML, which is gaining traction because most of the mobile devices use it for synchronization of data. I yet to find where CalDAV fits when it comes to mobile devices.

I found out some interesting links, which I am still going through. I thought, it would be nice to share these links: