[Beowulf] What is the right lubricant for computer rack sliding rails?
Robert G. Brown
rgb at phy.duke.edu
Fri Feb 6 09:15:10 EST 2009
On Thu, 5 Feb 2009, Geoffrey Jacobs wrote:
> > conducting powder. Personally I'd use WD, but hell, I'd cook
> with WD if
> > I couldn't find any olive oil...
> > (The main issue in any case is to be sparing and not spray it
> so it gets
> > sucked into cooling fans.)
> Also shy away from WD-40 as a general lubricant. It gets gummy over
> time. For slides look at a light grease, perhaps a white lithium
> grease like 3M™ White Grease commonly used on some garage doors and
> autos. It only takes a little....
Awww, y'all just don't understand. The issue is WHY are we lubricating
rack rails? As noted, a properly installed rack with brand new rails
that haven't been left out in the rain and have spent their life in a
sealed plastic bag or holding a piece of hardware up in a rack with
almost no actual movement should NEVER require lubricating -- really not
even the first time at manufacture. A high quality rail is basically
a double-hung track with e.g. some ball-bearings or rollers. In a
dust-free, moisture-free environment they won't rust and they won't gum.
So if your rack is "tight" and it is hard in ANY WAY to slide nodes or
servers in and out, it is almost certain that your rack is out of square
so that the rails are mechanically binding. They don't need
lubrication, they need for you to resquare your rack (if the problem is
persistent in a rack), or you need to replace the rail (in the case that
you have a single rail that is sticking) because THAT RAIL is defective.
Or perhaps you're using the wrong rails for your rack, mounting them
incorrectly, or something like that.
OK, so let's imagine a rail that is made out of extraordinarily cheap
steel, badly, so that brand new it does bind a bit or squeaks or
something. I don't think I've ever experienced a rail like that, but
I'm sure they must exist. You contact the vendor, and they're already
out of business, or have changed their business name so they can
continue to sell their military surplus rails that they found on ebay
that were stolen from a military compound in Viet Nam back in the 70's.
You've got a pile of them, and they all squeak and you can't replace
In that case, I would absolutely give every rail a shot of WD, run it in
and out a couple of times. Squeak goes away. Mount the nodes (making
sure rack isn't torquing the rails. Forget about it.
I mean, how often do you pull a node in its lifetime? 0.013 times (on
average)? If, for any reason, it squeaks when you pull the one node in
80 (presuming you didn't get ebay nodes as well) that maybe fails over
three or four years of service, hell, be extravagant. Give the rail
another squirt to shut it up (with the little tube-y thing so you don't
dump a lot of aerosolized WD in the air).
WD is cheap and readily available. Chances are you already have some
and don't even have to go buy it.
> There are also Teflon based white greases and many excellent but black
> and dirty molybdenum disulphide greases. Moly rich grease is
> in that the moly "plates" on the surface and if the tolerances are
> binding (not lubrication) can occur. On old cars however it can
> up things if used sparingly. As much as I like moly greases I think
> a multi-purpose lithium based white grease from an auto supply house
> the best choice in this case.
The jury is still out on whether or not Teflon is a carcinogen -- it is
a polymerized fluorocarbon. It is definely a bad, bad thing to have
around in any quantity if you exceed 500F (in a fire) and it starts to
decompose. Even unheated, it is worrisome. See e.g. here:
Some neutral "nonreactive" materials (like asbestos) have turned out to
be carcinogenic (perhaps the body manages to break down the materials in
ways similar to heat, hmmm). If it is, aerosolized microparticulate
Teflon is far more likely to be than an intact Teflon coating in a pan,
and you're delivering it to far more sensitive lung tissue than the
teflon you eat when you cook with teflon coated pans. I personally
don't buy teflon coated pans any more. Every one I've ever owned has
ultimately turned into a scraped off mess, and I know that some of the
scrapings have gone into me. If you use them to cook every day,
somebody can almost certainly TELL that you use them just by measuring
C-8 (a known carcinogen) levels in your blood.
> It only takes a little....
> Teflon grease can be purchased from Bicycle shops. It's apparently more
> durable than lithium grease over a reasonable temperature range -- less
> prone to go runny.
But look, this is a LOW, LOW, STRESS usage. It's like lubricating a
door that you're only going to open at MOST a half-dozen times in your
lifetime. To me a squeaky door means "reach for the WD", which usually
solves the problem for years if not indefinitely and doesn't leave a
But if you DON'T like WD and INSIST on lubing something that shouldn't
need to be lubed ever, I'd recommend using 3-in-1 oil (a.k.a. light
machine oil) delivered directly to the track if you can find a suitable
lubrication or access point in a device designed to never need to be
lubricated (again) after manufacture. This kind of oil is basically s
tiny bit thicker than WD, doesn't spray, does drip. It takes longer to
evaporate. Like WD, it works wonders on hinges or gummed-up drawer
rails -- like ANY oil it will attract dust and turn into gum itself if
used in a dirty environment (which shouldn't be a problem in a machine
room). It is probably very similar to what the manufacturer applied (if
they applied anything) at the time of assembly, although they might have
used a light grease instead of a light oil if the roller assembly is
mostly closed -- grease is even worse than oil about attracting dust and
gumming up as it starts a whole lot more viscous to begin with. That's
why greased bearings are usually sealed, or one expects to perform
regular cleaning and regreasing maintenance on e.g. bicycle chains
(where WD is again a lovely quick fix that cleans AND lubricates, even
if you add a touch of grease afterwards because the WD will evaporate
off too quickly).
As for volatility, fire hazard, etc. -- a total straw man. A drop or
two of WD or any other light oil in the rails isn't going to create a
fire or explosion hazard. There's probably a drop or two there already
from manufacture. Saturating the rails and using up a whole can of
anything per rack is obviously stupid, and even THAT would be hazardous
for only a few hours after application -- the air in a server room is
constantly being moved and changed over; you're not pooling up a gradual
release to explosive levels like you might if you dumped a jar of
gasoline out on the floor of a sealed unventilated room and walked away.
People use WD and light oil in hot sparking machine SHOP rooms all the
time and don't think twice about it.
OK, that's more than enough prose devoted to something that nobody
should ever have to actually do in any machine room unless they're using
rails they inherited from their grandfather that were stored in the
basement for twenty years OR failed to square up their racks properly
when assembling them so that they are binding. In which case they still
shouldn't -- they should square up their racks.
Seriously -- can anyone on this list document a single case where they
had to lubricate rails to get them to work (where it wasn't immediately
apparent that the rails were defective and really needed to be returned,
not lubricated)? Or where they did anything more than reach for
whatever lube they had handy if they did, once, have a pair that was
troublesome and had to be lubricated to get them to work at all?
> MORE CORE AVAILABLE, BUT NOT FOR YOU
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
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
-------------- next part --------------
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
More information about the Beowulf