Bug#756521: ITP: kadeploy -- Scalable, efficient and reliable cluster provisioning solution
On 02/08/14 at 15:38 +0800, Thomas Goirand wrote:
> On 07/31/2014 12:02 AM, Lucas Nussbaum wrote:
> > Package: wnpp
> > Severity: wishlist
> > Owner: Lucas Nussbaum <lucas@debian.org>
> >
> > * Package name : kadeploy
> > Version : 3.3
> > Upstream Author : Kadeploy developers <kadeploy3-devel@lists.gforge.inria.fr>
> > * URL : http://kadeploy3.gforge.inria.fr/
> > * License : CeCILL version 2.0
> > Programming Lang: Ruby
> > Description : Scalable, efficient and reliable cluster provisioning solution
> >
> > Kadeploy is a scalable, efficient and reliable deployment system (cluster
> > provisioning solution) for cluster and grid computing. It provides a set of
> > tools for cloning, configuring (post installation) and managing cluster nodes.
> > It can deploy a 300-nodes cluster in a few minutes, and also supports
> > authorizing users to initiate their own nodes deployments (including with
> > concurrent deployments).
> >
> > A work-in-progress package is available from:
> > Vcs-Git: git://scm.gforge.inria.fr/kadeploy3/kadeploy3.git
> > Vcs-Browser: https://gforge.inria.fr/scm/browser.php?group_id=2026
> >
> > Lucas
>
> Have you compared it to:
> https://github.com/enovance/edeploy
>
> which has nice role-based system, so you can deploy specific systems
> depending on what type of hardware (amount of RAM, number of HDD, or
> anything else you decide)...
Hi,
I don't remember the exact details, but I think that edeploy's design
made it scale poorly (especially for the image broadcast part). This
might not be a problem for the typical use case for edeploy (one-time
deployment of small/medium-scale clusters to build an OpenStack
infrastructure), but it would definitely be a problem for the typical
use case for Kadeploy ((re)install medium to large scale HPC clusters
during maintenances). Cloning 112 nodes with a standard Debian image
with Kadeploy takes about 5.5 minutes, with most of it spent waiting for
the nodes to reboot. We are working on using Kexec to make reboots
faster, but unfortunately there are quite a lot lof bugs to work around
there.
Lucas
Reply to: