[Beowulf] motherboards for diskless nodes
rsweet at aoes.com
Sat Feb 26 02:35:37 PST 2005
Thanks for the injection of some perspective. It is clear that I chose my
phrases poorly, and some alternate wording would have been more constructive
On Fri, 25 Feb 2005, Greg Lindahl wrote:
> On Fri, Feb 25, 2005 at 06:52:59PM +0100, Ryan Sweet wrote:
>> While I can understand debating over the merits of nfsroot vs RAM-disk
>> root, I fail to see many useful arguments for maintaining a local OS
> An example of something that goes very wrong with NFS is upgrading a
> file to a new file with the same name. If that file is a binary or
> library that's in use anywhere in the cluster, you are likely to have
> a problem. Local disks and Scyld, on the other hand, do the right
> thing: existing processes using the binary or library continue to use
> the old version, while new ones use the new version.
In practice I think the importance of this depends upon the particular
requirements of the site. Definitely there must be some places where its
important, and should be planned for. However most sites I've seen accept
(prefer) that new versions of application software be installed alongside the
old, rather than in replacement of it, and for system software updates they
are usually willing to accept stopping running jobs and starting them again
(either with the queue system or without.
> This disagreement is as old as the hills, by the way: in the good old
> days, when Sun was young, lots of people ran their pizza-box
> workstations diskless, but that went out of style when Ethernet's
> performance was stuck in place for a bunch of years.
Or in some places it kept right on going.
> It's important to understand arguments you disagree with; your
> dismissal is not a good sign.
> Yep. But your conclusion:
> Doesn't make any sense; I have seen people describe such systems where
> they download a disk image when a batch job wants a different software
> load. It's certainly doable that way: it does have different tradeoffs
> from the diskless case, but if it gives you a headache, it's probably
> because you don't like it, not because it's hard to do.
Yes, I agree. I chose my words poorly in the first post, and came down rather
hard against local installs.
In the end the choice is about balancing managing complexity with the
requirements of the particular site. I think that in a large percentage of
use cases the admin will find that managing "diskless" (local disk for
swap/scratch) systems is highly advantageous.
Ryan Sweet <ryan.sweet at aoes.com>
Advanced Operations and Engineering Services
AOES Group BV http://www.aoes.com
Phone +31(0)71 5795521 Fax +31(0)71572 1277
More information about the Beowulf