2008/7/11 Etienne Goyer <[EMAIL PROTECTED]>: > We (some colleagues and myself) are currently working on white paper > about HA clustering using redhat-cluster-suite on Ubuntu. We would like > to hear about different real-world deployments to be able to synthesise > a set of best practices. The cookie-cutter Apache/NFS/Samba scenarios
At Brightbox we currently use the cluster suite with our virtualised Ubuntu Ruby on Rails service. We only use a bit of it, and essentially we have found it to be more trouble than it is worth. A great number of our availability problems are due to problems with the cluster technology not behaving properly. Essentially the issue is that the cluster suite couples machines closely together - that is its purpose. And like all coupling it creates a system that is brittle in the face of unforeseen problems. So we've taken the view that we're going to eliminate both clustering and NFS from our setup and instead use GlusterFS (http://gluster.org) and database clustering as our shared facilities. The rest of the design is going to be shared nothing as far as is humanly possible. Anybody who has struggled with clustering, NFS and yowled with frustration at stale file handles will find GlusterFS a complete breath of fresh air. There are Ubuntu Hardy packages for GlusterFS here: https://edge.launchpad.net/~ubuntu-ruby-backports/+archive for anybody who fancies giving it a whirl. -- Neil Wilson -- ubuntu-server mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-server More info: https://wiki.ubuntu.com/ServerTeam
