Installing a Docker Swarm cluster inside VirtualBox with Docker Machine

I’ve documented the process of installing a Docker Swarm cluster inside VirtualBox with Docker Machine. This allows experimenting with Docker Swarm, the simple docker container orchestrator, over VirtualBox.

This allows you to play with orchestration scenarii without having to install docker on real machines.

Also, such an environment may be handy for teaching if you don’t want to install docker on the lab’s host. Installing the docker engine on Linux hosts for unprivileged users requires some care (refer to docs about securing Docker), as the default configuration may allow learners to easily gain root privileges (which may or not be desired).

See more at

Making Debian stable/jessie images for OpenStack with bootstrap-vz and cloud-init

I’m investigating the creation of VM images for different virtualisation solutions.

Among the target platforms is a destop as a service platform based on an OpenStack public cloud.

We’ve been working with bootstrap-vz for creating VMs for Vagrant+VirtualBox so I wanted to test its use for OpenStack.

There are already pre-made images available, including official Debian ones, but I like to be able to re-create things instead of depending on some external magic (which also means to be able to optimize, customize and avoid potential MitM, of course).

It appears that bootstrap-vz can be used with cloud-init provided that some bits of config are specified.

In particular the cloud_init plugin of bootstrap-vz requires a metadata_source set to “NoCloud, ConfigDrive, OpenStack, Ec2“. Note we explicitely spell it ‘OpenStack‘ and not ‘Openstack‘ as was mistakenly done in the default Debian cloud images (see

The following snippet of manifest provides the necessary bits :

name: debian-{system.release}-{system.architecture}-{%Y}{%m}{%d}
  name: kvm
  - virtio_pci
  - virtio_blk
  workspace: /target
  # create or reuse a tarball of packages
  tarball: true
  release: jessie
  architecture: amd64
  bootloader: grub
  charmap: UTF-8
  locale: en_US
  timezone: UTC
  backing: raw
    #type: gpt
    type: msdos
      filesystem: ext4
      size: 4GiB
      size: 512MiB
  # change if another mirror is closer
    password: whatever
    username: debian
    # Note we explicitely spell it 'OpenStack' and not 'Openstack' as done in the default Debian cloud images (see
    metadata_sources: NoCloud, ConfigDrive, OpenStack, Ec2
  # admin_user:
  #   username: Administrator
  #   password: Whatever
    # reduce the size by around 250 Mb
    zerofree: true

I’ve tested this with the bootstrap-vz version in stretch/testing (0.9.10+20170110git-1) for creating jessie/stable image, which were booted on the OVH OpenStack public cloud. YMMV.

Hope this helps

Handling video files produced for a MOOC on Windows with git and git-annex

This post is intended to document some elements of workflow that I’ve setup to manage videos produced for a MOOC, where different colleagues work collaboratively on a set of video sequences, in a remote way.

We are a team of several schools working on the same course, and we have an incremental process, so we need some collaboration over a quite long period of many remote authors, over a set of video sequences.

We’re probably going to review some of the videos and make changes, so we need to monitor changes, and submit versions to colleagues on remote sites so they can criticize and get later edits. We may have more that one site doing video production. Thus we need to share videos along the flow of production, editing and revision of the course contents, in a way that is manageable by power users (we’re all computer scientists, used to SVN or Git).

I’ve decided to start an experiment with Git and Git-Annex to try and manage the videos like we use to do for slides sources in LaTeX. Obviously the main issue is that videos are big files, demanding in storage space and bandwidth for transfers.

Continue reading “Handling video files produced for a MOOC on Windows with git and git-annex”

A howto record a screencast on Linux and tablet

I’ve documented the process of how I’m trying to perform DIY screencast recording, for the needs of a MOOC.

I’m working on my Debian or Ubuntu desktop, using an external graphic tablet with integrated display for annotating slides.

The main software used for the process are xournal for annotating PDFs and vokoscreen for the screen and video recording.

Here is the documentation :

And here’s the companion video :

I hope this is useful to some.

How to publish an HTML5+RDFa Web site from org-mode

I’m a big fan of org-mode (see previous posts), and I’ve started maintaining (sic) my professional webpage(s) with it.

But I’ve also recently tried and publish some more Semantic/Linked Data aware documents too (again, previous posts).

Ideally, I think my preferred workflow for publishing articles or documents of some importance, would be to author them in org-mode, and then publish them as HTML5 including RDFa meta-data and annotations. Instead, I’ve more frequently been doing conversions of org-mode to LaTeX, in order to submit a printable version, and later-on decided to convert the LaTeX to HTML5+RDFa…

But one of the issues is how to properly embed the RDF meta-data inside the org-mode documents, so that the syntax is both compact and expressive enough.

I doubt there’s a universal solution, given that RDF tends to be complex, and graphs may not project easilly along a mainly linear structure of an org-mode document, but anyway, there seems to be possible middle grounds that are practically good enough.

I’ve tried and implement a solution, which reuses the principles set by John Kitchin in Extending the org-mode link syntax with attributes, i.e. implementing an HTML exporter for a particular custom link type, which will convert the plist-like syntax to some RDFa constructs.

Here’s a description of the whole solution :

The nice thing about org-mode, and its litterate programming babel environment, is that it allows to embed the code of the links exporter inside the org document, avoiding to dissociate the converter from the document’s source, making it auto-complete.

Next step will probably be to author a paper (or convert back a “preprint” of mines) with org-mode, in order to provide Linked Research meta-data.

Stay tuned for more details, and in the meantime, I welcome any improvement to the org/babel/elisp setup.

Edit: I’ve recorded a webcast to provide a bit more details, available on YouTube :