During these Christmas Holidays, I’ve decided to migrate this blog from Drupal to Octopress (after 2 years of goodwill).

Octopress is quite popular among opensource developers since its include all the features needed to maintain a simple blog plus some extensions like source code syntax hilight, social networks integrations and support to deploy on GitHub Pages and Heroku (both free).

I should say that Octopress isn’t a CMS as many of you can imagine, but a static html pages generator. This may sounds oldie, but Octopress leverage some existing tools, and you will like it.

The following is a Google Webmaster graph, the performance gain was from an average of 600 ms for page retrivial with Drupal (with caches) to an average of 120 ms with Octopress.

To migrate existing posts from Drupal 7, I’ve used a slightly modified version of an existing migration script. Put your mysql connection details and execute it, it will generate a bunch of files to put inside the source/_posts folder of your Octopress.

These months have been quite busy for me, but I always try to find some time to seed on local communities the knowledge of some particular Open Source software, especially for OpenNebula (someone called me evangelist, but it is really too much for what am I doing).

On 26 Novembre 2012 I carried out a workshop, sponsored by Develer (offering space, organization and drinks), that attracted quite many people around the Florence area, spending a nice evening discussing about Cloud Computing with OpenNebula.

The presentation covered some basic topics, like what OpenNebula is and why users should use it, and continued with a live demo using the OpenNebula Sandbox VM.

Are you wondering how many alternative ways to get notifications from your nagios/icinga exists, other than plain email messages? And what if you actually are monitoring your email or DNS infrastructure and you don’t want to use a third-party email account dedicated to our precious notices?

An alternative and effective method to deliver the short status notification is via Twitter messages, using a protected account (for privacy).

During these first hot-days in Florence, I’ve attended the EVA Florence 2012 conference, presenting a talk with Cristiano Corsani about a project for the National Library of Florence (BNCF), covering the deployment of a small Private Cloud using exclusively open source software, for serving internal and public BNCF services.

The proposed solution includes a virtualization stack running on commody hardware, using the Linux KVM hypervisor, managed by Proxmox 2.0 and with a GlusterFS shared storage.

I hope to write down some article to share the most interesting bits of this configuration, but time is not our friend.

Here are the slides of the presentation “Open low-cost HA cluster cloud”, hoping that they will satisfy your curiosity:

Thanks to everyone who helped us in this journey!

During the migration of a RAID1 mdadm on Ubuntu server to a ZFS mirror, I got stuck at the initial zfs volume creation with the following error:

cannot create ‘tank’: one or more vdevs refer to the same device, or one of the devices is part of an active md or lvm device

Obviously I’ve removed the device from the old mdadm array with:

1
2

sudo mdadm /dev/md3 --set-faulty /dev/sdb3
sudo mdadm /dev/md3 --remove /dev/sdb3

However, the zpool utility doesn’t actually look if the device is currently used or not, but simply look at the metadata at the beginning of the disk that is still there.

To wipe the mdadm metadata you need to use:

1

sudo mdadm --zero-superblock /dev/sdb3

But the good old dd will do the job too:

1

sudo dd if=/dev/zero of=/dev/sdb3 count=64 bs=1024