Re: [SSI-users] How do I make every node have direct access to the file system?

[Christopher G. S. I. <cg...@ld...>]

----- "CooLCaT@Vienna" <co...@co...> wrote:

> i am trying following apporoach:
>
> the root file system resides on a nfs share. every node has a
harddisk
> with
> only the boot partition on it. we want to achieve, that every slave
> node can
> become master node so that all nodes must die before cluster gets
> offline.
>
> any better ideas somebody?

That's a pretty tall order.  You would need to implement a new quorum
system with something like Paxos, as well as all of the associated
machinery (mirroring, syncing, etc.) to go along with NFS.  Maybe
GlusterFS or some wonky mix of OCFS2, drbd, Heartbeat, and a lot of
scripting would work.  Heck, you could even stuff a filesystem into
MySQL Cluster, but it won't be very much fun. :)  You're better off
with the aforementioned GFS or Lustre.  The easiest route will be
evaluating whether or not you really need a shared writable partition
at all.  A read-only NFS root partition share mounted as CoW in RAM
(or just some creating symlinking) will take care of things for most
situations, except for that pesky inability to propagate writes
without a reboot of every node.

-- 
Christopher G. Stach II