CoreOS is the shooting star among cloud operating systems
Fleet
Starting and stopping services on a cluster with Pacemaker is often more trouble than just doing it manually. For a relatively simple task, many admins would prefer to avoid the complexity of a tool like Pacemaker. For a simpler approach, the CoreOS developers wrote an extension for systemd, which they call Fleet.
Fleet adds awareness of cluster tasks to systemd: the Fleet instances on all nodes talk to each other and then run specific commands on the local systemd instances. In combination with etcd, Fleet ultimately allows centralized control of a full cluster.
Release Plan
The way the CoreOS developers plan their releases seems a bit confused at first glance, but on closer inspection, you can see three release branches: the alpha branch, in which all the latest developments take place, the beta branch, in which the tested and the next release are prepared, and the stable branch that contains the current stable version.
The plan for releases is strikingly similar to the Debian release cycle, and the methods are also similar. For example, updates migrate from the alpha to the beta branch after a certain period of time, and if no new bugs are identified, the beta becomes the new stable tree. (The CoreOS developers go through the whole process a bit faster than the Debian community.)
If you want to try out CoreOS, you are well advised to go for the beta version at least. For initial experiments, the latest stable version might even be preferable.
The CoreOS site [7] provides a variety of information on how to install CoreOS. Testing with Qemu, Libvirt, or VMware is highly appreciated and technically easy to implement.
Online Resources Help with Testing
For a test, the variant based on an OpenStack installation is a good choice – the corresponding OpenStack image is even provided on the project website.
Using a script, you can pass the discovery IDs to the etcd instances on the VMs. After doing so, you will have a multiple-node cluster of CoreOS VMs that are orchestrated via etcd and Fleet. After that, you could basically already start on the Docker containers, in which you then run your applications as containerized services. In our lab, CoreOS always worked exactly as the developers promised, and it showed no weaknesses – many kudos!
« Previous 1 2 3 Next »
Buy this article as PDF
(incl. VAT)