started at a new customer we were looking for a more flexible way of having jenkins spinning up slaves on the fly. This in a way a slave is only started and consuming resources when a specific job is running. That way those resources could be used more efficient.

Also the fact that developers could take control over their build servers by managing the Dockerfiles themselves is a great advantage too. But that's for a later phase. Let's start at the beginning.

For the docker host a CentOS 7 server has been provisioned and prepared to run the docker daemon …

A couple of years ago I deployed an openstack cluster based on RDO. Back in the days we implemented the kilo release. Until today we didn't updated yet due to various reasons being no need for the new features, no resources, no time no.. Upgrading would be a better option but we'll have to live with it and since it's running rather well so far we are quite happy with it.

To manage that cloud I use the clients I installed on my local machine, from nova to cinder they all have different packages available for many different platforms. Only …

many years ago I created my own webpage, it all started with pure, HTML evolved to a wordpress and finally became a pelican based setup. It got served on many different hosting providers but since a few years it's running on S3 storage and hosted through cloudfront all over the world.

It's a very fast setup, and once the site has been deployed and every little service has been configured and implemented the only thing I need to do is writing content in markdown without having to consider how to deploy or how it will look.

In this post I'll …

Recently I was been asked to increase the MTU on the deployed openstack cluster at one of our customers. Since the beginning of my touch on openstack networking has been the hardest part to get my head around. In the first place because openstack does some nifty things on the networking path. But also cause for the use case at the customer a lot of customization has been done to get it implemented in their infrastructure.

Hence the shiver when the MTU question was been made..

Nevertheless together with a colleague who likes a challenge and has a profound knowledge …

Some weeks ago an article on hacker news got my interest. From time to time I really get an healthy dose of jealousy when people found an idea which could make them buy a tesla. My terms of someone who make a lot of money ;)

This one is so brilliant in it's simplicity that I really was flabbergasted and made me wonder why I never came up with the idea. It generates nice reports of the usage of your site which is hosted by aws. Based on the logs of the S3 bucket or the cloudfront domain you setted up …

for one of my projects, the sms-twitter wall setup, I configured a raspberry pi with 2 wireless network interfaces to connect through a hotspot enabled on an android device. I discovered on previous events that the wireless adapter failed on me from time to time. So I went to the internet to look if I could add a second interface and bond them together.

I found a lot of documentation on how to bond an active-backup strategy with a wired and a wireless interfaces but didn't found a setup with 2 wireless interfaces. After a while I figured out how …

About a year ago I attended the AWSome Day at Mechelen. Back then I wrote a first draft article about it, but it got out of my sight unfortunately. I reviewed it and decided to publish it anyway.

The event was based on their essentials course and took use through the different AWS core services (compute, storage, database and network).

I do know it has nothing to see with open-source. But it is a part of that ultimate cloud based setup I believe in which exists in one central place from where you can manage all your virtual machines independent …

last couple of days I have been fighting with the way an static ip is configured on an openstack virtual centos 6 instance. In our specific use case we ditched as many network openstack services as possible as I previously described.

We want to have the instances running in our current network spaces of the R&D department. In this department until some days ago we didn't had any DHCP server running. But a few weeks back we added an extra remote network space into our platform where we configured a remote compute-node.

This is where the issues started popping …

Some of you may already have notices others just stumbled on this post through a search engine, I have set up an openstack private cloud at one of our projects:

• vlan flat-neutron provider network
• layer2

We have noticed that the benefits of having a private cloud is spreading through the different teams within the organization and therefore the interest into this flexibility is growing. Since this wasn't the original use case we are encountering some design issues right now.

For the original instances the default overcommit ratios are fine. But the request for new machines with other goals are like …

A few months ago I implemented an RDO based openstack kilo release private cloud at one of our customers for their development platform. Through time we tackled a couple of issues so the cloud could be fitted into their work flows.

We stumbled onto some minor issues and some major ones. Let's begin with the minor ones ;)

When upgrading the all-in-one controller before we started using the cloud in 'production' a mean bug bit us in the ankle due to a new hiera package. After some digging around a patch came to the rescue together with the exclusion of the …