math at velocet.ca
Tue Jan 14 10:21:20 PST 2003
On Tue, Jan 14, 2003 at 12:47:50PM -0500, Robert G. Brown's all...
> > Without metal you get way too much RF. We found that stacking nodes
> > really close together vertically with no shielding caused many to not
> > boot properly (seemingly the most sensitive time) and many to crash
> > randomly (perhaps 10% of nodes every 2 weeks, mysterious crash).
If you think about this, its *VERY* hard to boot nodes, as it all had
to be done from the *CENTRE OUTWARDS*.
Then once a middle one crashed, it would take 10-15 tries to get it to
boot up properly if it was in the centre. A few never seemed to boot ever
again unless a few nodes around them were also off. We took that configuration
apart after a few weeks of frustration. ;)
> didn't have my entire collection of music on my hard disk. If I was
> tuned to anywhere in the vicinity of 100 MHz while the PC was running,
> the radio buzzed like crazy. Don't ask me why -- it had a 4.77 MHz
> clock, IIRC, so it must have been harmonics of some sort. I could
> clearly hear the computer "work" -- the buzz was nicely modulated when I
> ran a small floating point program, for example. Made me truly believe
> that someone could read your keystrokes off your power supply line or
> local system EMF with sufficiently delicate equipment a la
Nowadays, you dont need a radio, you just need a cellphone.
Both my laptop speakers on my 600Mhz P3 laptop and my cell phone will buzz
like nuts when one or the other is near. The PC buzzes in bursts, slow
bursts when im not talking and quick ones when I am or the other person is.
When the laptop is idle, it buzzes in quick patterns, if I do a fair bit
of stuff on it (compile, move windows, surf), I just hear solid buzzing --
probably the memory refreshing?
> At a guess you could have also solved your problem by inserting e.g.
> aluminum screens with mesh holes on the order of a few mm in between
> nodes. Depending on whether or not you got reflection and self-induced
> crash or absorption.
What 'absorbs' RF? (other than stealth bomber paint)
> However, as I noted to one person off list, the things that have
> prevented me from ever building a cluster like this are:
> a) Liability, fire codes, personal safety, and (as noted) emf. In many
> institutions you'd get flayed alive installing electrical hardware in
> non-UL-or-other-safety-underwriter-underwritten configurations. A small
> "lab" or "hobby" configuration maybe -- in a large scale production
> facility no no no. Just establishing predictable airflow cooling and
> controlling dust argue against an open large scale design.
This was a small 'hobby' installation to test wether this was the gear
we wanted. 12 nodes only I think, at the time. The cabinet we had
designed wasnt ready yet. (The cabinet worked much better of course,
being solid plate steel and easily grounded.)
> b) Time, beauty, and convenience. Time because it takes a lot less to
> just use OTC parts. Beauty matters to e.g. granting agencies -- they
> are more inclined to think a cluster is a "real" resource if it looks
> like a rackmount or even a tower/shelfmount beowulfish computer than if
> it looks like my latest science project with everything out naked and
> wires exposed. Convenience (really more time) because if a node fails,
Our cabinet in the end with its smoked glass door with its futuristic
rivetting/bolting pattern (no charge) looked *MUCH* nicer than a tower case.
People see towers and go "hmm regular computer, possibly a low end server".
When they see the Monolith (as it got dubbed) they're much more impressed ;)
> it is encapsulated and can be benched to work on it. Unbolting a power
> supply from a shelf of running units, pulling the motherboard out of a
> shelf of running motherboards? Messing with cards ditto? Adding a hard
> disk to a shelf-mounted system? Standardization saves time, and cases
> are reasonably standardized.
This is why we had a custom cabinet with appropriate mounting braze-ons for
screwing in motherboards, mounting power supplies and even a 3.5" mini
mounting L bracket in case they wnated to add hardisks later to the
> Custom stuff pushes you out of the OTC/OTS sweet spot, into greater
> risk, less explored engineering terrain, uglier and less
> professional-looking installations, and ultimately can cost time and
> money beyond what you expect to save. I'd avoid them unless you are a
> devoted hobbiest and/or have some very specific engineering constraint
> that can't be met with OTC parts and configurations.
I wanted it to look like a cluster cabinet, not a server PC (or rack thereof).
The cabinet looks way nicer than that ;) As I said, the cost was about
the same as PC cases in the end (and we even had a cost overrun -- it
would have been cheaper than cases, had I planned a little more carefully,
but I am pretty happy with my first ever attempt at a custom cabinet (the
actual manufacturers had much experience with building such things, however,
they could not have forseen the way we were going to use it once it
was in place (I didnt either, which was the problem). No major hassles have
resulted, so they're minor and quite easy to fix for next time.
> With that said, I >>have<< had occasional fantasies about doing a
> home/hobby shelf of nodes this way. There my time is "free", my space
> is very limited. OTOH, I burn myself, my family, and my pets alive if
> it starts a fire in the middle of the night that gets out of hand too
> quickly, or maybe just lose everything I own. And presumably safe(r)
> tower cases that might possibly confine a the heat/fire from an internal
> short until a fuse blows are so cheap. And my nodes accumulate dust
> (see "pets") so fast even when closed...
How many clusters have gone up in flames? :) Not that this is a concern,
it just suprises me that things could get that hot.
Arent things generally considered ok if the high voltage (AC 110) is in
the powersupply box? The boards are using 12 and 5v and lower, which
isnt regulated nearly as tightly. (Our cabinet was made of metal so
im not too worried here, just asking for future 'build a test cluster
on a shelf' designs that I am *SURE* _MANY_ beowulfers are not
admitting to having running AS WE SPEAK! :)
> Robert G. Brown http://www.phy.duke.edu/~rgb/
> Duke University Dept. of Physics, Box 90305
> Durham, N.C. 27708-0305
> Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
Ken Chase, math at velocet.ca * Velocet Communications Inc. * Toronto, CANADA
More information about the Beowulf