| Friday, October
20, 2000
Dear Dr. Pournelle:
In Friday's VIEW, you opined:
> Moral: Intel and 3Com hubs cost about 3
times as much as Netgear. > They are also more reliable. Reliable is a
relative term: this is the > first failure of any Netgear hub in years,
and we use many of them. > It cost me about an hour of my time. I would
have saved money by > buying a 3Com, as it happens, except that in my
case this gave me > something to write about. You probably won't have
that upside.
I have about 200 ports here. In the seven years
since we changed from ARCNET, have changed form concentrators to switches
to faster switches, so let's say I've had about 600 ports to oversee.
Using the 3Com gear you cite (and spendier versions than most users would
buy), I've had four ports on a variety of concentrators fail, plus on
entire switch go dead (a month past end-of-warranty).
>From my sample, that's a failure rate of .006;
how does that stack up to the Netgear, failures-vs-samples?
FYI, I'm in a building with large and constant loads
in the downtown core of an area with very clean power (the Pacific NW has
the lowest surge-and-spike rate in the nation as per a Nat'l. Power
Laboratory Power Quality Study quoted by APC), and EVERYHTING runs on an
APC Net7 or Net7T surge protector (rated at <40v passthrough during a
spike or surge using the UL-1449 standard test). I suspect your
residential area up on a ridge or crest (I've seen your pix of walks w/
Sasha) does not have the same steady power as my office does.
Point I'm trying to make is the 3Com gear is by no
means infallible, even when used by professionals with precautions in a
relatively safer environment. I doubt that 3Com would have saved you any
time, even billing at what you are worth ($300/hr as a consultant?).
--
Very happy that Phillip was not on the COLE, and had
thought of him when I heard the news, as it slipped my mind what ship he
was on.
Thank you kindly.
-- John Bartley, PC syadmin, USBC/DO, Portland OR
Views expressed herein are mine own. "LOS ANGELES: A city of
millions; thousands more are born each day. Some in maternity wards, some
in creche incubators. The Artificial ones don't have civil rights, but
they still need the law. That's why they turn to me. My name is Friday. I
carry a badge." -- Robert A. Heinlein's "Dragnet"
<<Courtesy Ray Radelin>>
Good points. I probably ought to rethink the
situation. In any case I have put tape over that one bad port and will
continue to use the Netgear concentrator.
Dr. Pournelle,
Regarding power failures: I had a similar problem
with a computer owned by a human-service organization which is housed in
the church where I work. This organization is not connected to our LAN and
has only one computer.
After a power failure, they couldn't get their
machine to do anything except show the "on" indicator light. I
wound up getting the machine to work by unplugging it. The only thing that
makes sense to me is some type of power management glitch.
All this power management reminds me of the '70s,
when college libraries shut off their copiers "to save
energy"--it should have been obvious to the veriest dolt that
frequent power-cycling would increase problems with copiers that student
treated none too gently anyway. One additional service call per month
probably caused the service technician's truck to burn more energy just
driving there than than any campus saved by having the copiers turned off.
Maybe a case could be made to the EPA that spammers
are wasting energy and causing pollution?
jomath
Hi, Jerry. I believe that on many Intel 810 and 815
motherboards, holding down the power switch for 4 seconds will do a hard
powerup or powerdown.
--Mike Fessler
I believe that too, but it sure didn't work.
Dr. Pournelle, I have worked on three systems this
year, that I can recall, with the same problem: dead, cycling the switch
doesn't bring it back. Call the tech (me). Well, for the first two, when
we got the pc back to our shop to check it out, it worked fine. We noticed
the coincidence. All different manufacturers, but, all Pentium II or III,
ATX boards. Somehow or another, some sort of static charge or such must be
binding up the power supplies, causing them not to switch on. Several
minutes later, more than 10, and after unplugging the system from the
wall, it will dissipate, and begin to work. Not yet ready for prime time
technology, if you ask me. I haven't noticed whether or not an AMD based
system will do the same, though I imagine so, it seems to be rooted in the
ATX power supply. And if I am not loosing my mind, I seem to recall that
all the power supplies were Enlight. Yours too? I use Pournelle's Law
religiously, 90% of problems are with the cable, but perhaps we should
create a new law governing ATX power supplies? If the machine is dead,
unplug it for half an hour and try again.
Just a note to let you know you aren't the only one
cursed by temperamental machines.
George Laiacona III <george@eisainc.com> ICQ
37042478/ 28885038 "Son, when you participate in sporting events,
it's not whether you win or lose, it's how drunk you get." -Homer
Simpson
Hi Jerry,
The problem you describe with your i815/PIII-933
happens occasionally with ATX power supplies. Not all brands seem to do
it, but whenever we have a power failure at work invariably there are a
half-dozen or so systems that don't come back up, so we run around pulling
plugs, waiting 10 seconds, then plugging back in and powering on. An ATX
computer that's powered off is never truly powered off unless it's
unplugged--you can get an indication of this by watching the Wake-on-LAN
lights on certain network cards. Plug a recent 3Com NIC into a LAN, and it
lights up even if the computer's not on.
As for hubs, I concur with you that Bay Netgear hubs
are pretty reliable. How much less reliable than a 3Com or an Intel, I'm
not sure. I've seen a number of 3Com hubs with one or two bad ports, but
the rest of the hub is still OK, so we just tape over that port to keep
from using it again. I don't have any experience with Intel hubs. I
suspect 3Coms and Intels may be more reliable because they do cost more
but without a comparable sample size of Netgears to compare it's only a
suspicion.
I don't hesitate to buy Netgear stuff, but I will
say I avoid anything cheaper than Netgear. I came home one day to find my
NDC hub totally dead for no explicable reason, and I had a cheap D-Link
NE2000 clone NIC die out of the blue after about a year of service. I
don't find $15 NICs to be worth the trouble. And with 3Com and Intel cards
sometimes selling for $35 or $40, I really wonder if my next NIC won't be
a tier-one brand, seeing as that's what I was paying for Netgears a year
or so ago and less than I was paying for no-name cards two years ago.
-- Dave Farquhar Author, Editor and Systems Analyst
Mail: dfarq@swbell.net Web: www.access2k1.net/users/farquhar
Dr. Pournelle -
Thank you for all you do, and trying things so we
don't have to.
"A power failure that lasts longer than the UPS
was a pretty serious test."
While preparing for the possible Y2K events, I
thought through this issue. Most UPS units only have enough battery for 15
or 20 minutes, but you don't have to accept that. My main UPS is now
equipped with a large external deep cycle battery from Sam's Club which
will keep it running for many hours (estimated at 1/2 day if all computers
are on). Wireless phones, DSL Router and Firewall/switch are also on this
power, but not the printer or scanner. The 12 volt power connections are
on quick disconnects wired according to the ham radio common practice, and
arraigned so I can switch batteries by plugging in the second battery
before disconnecting the first.
My furnace is equipped with a manual changeover to
an inverter run by a couple of golf cart batteries. This did not require
the more expensive UPS since no one will freeze, nor will anything be
damaged if power is out for a few hours. The batteries will run the
furnace for 24 to 36 hours depending on how cold it is out. This was
tested by killing the power and seeing how long I could run.
If the power is out for more than a few hours, I can
drag out the generator. I bought a large battery charger which will bring
all the deep cycle batteries back up to full charge from dead in no more
than 5 hours of running the generator. Gas is treated with stabilizer and
stored protected from the weather but outdoors away from the house. Marine
grade Amsoil synthetic oil is in the generator.
What you reminded me to do before this winter is:
1) check the acid level in all the deep cycles and
top them with distilled water 2) check the charge state in each cell 3)
charge the batteries as needed 4) change the oil in the generator 5) run
the generator for an hour under load 6) dump the stored gas into the auto
tanks and put fresh gas with stabilizer into storage 7) run another drill
with power off to confirm that everything still runs as long as I expect
8) recharge the batteries after the drill :>) 9) test all the radios,
AM/FM/SW, family radio band, ham, CB
None of this is very expensive to do (except buying
the generator - which also has other uses). I hope others may get some
ideas from this.
Ray Rayburn Audio (at) Technologist (dot) com
Jerry,
I can't verify the accuracy of this, but it sounds
authentic to me.
BTW, I'm really glad your son wasn't on this ship -
as I know you are, too. My first cousin was in this area on another ship
just last year.
A lot of families are giving thanks for those who
weren't there, even as we pray for those that weren't so fortunate.
JA -- John Alexander Faculty Resource Center The
University of Alabama johnalex@bama.ua.edu
---------- Forwarded message:
My brother is a civilian employee with the
Navy and sent this to me. Toni
I just got an e-mail from a good friend of
mine who flys SH-60B's off the USS Hawes. I thought you might like to read
it.
"It wasn't until a few days ago though,
that we started doing something that I feel may be the first thing I've
seen in my short Naval career that has truly made a difference. Right now
we're supporting the USS COLE and her crew in Aden. When the attack
occurred we were a day away. Just by luck we happened to be on our way out
of the Gulf and headed towards the Suez and could get here in a relatively
short amount of time. I know what you all have seen on CNN, because we
have seen it too.
I just want you all to know that what you see
doesn't even scratch the surface. I'm not going to get into it for obvious
reasons. But I will tell you that right now there are 250+ sailors just a
few miles away living in hell on Earth. I'm sitting in a nice air
conditioned state room, they're sleeping out on the decks at night. You
can't even imagine the conditions they're living in, and yet they are
still fighting 24 hours a day to save their ship and free the bodies of
those still trapped and send them home. As bad as it is, they're doing an
incredible job.
The very fact that these people are still
functioning is beyond my comprehension. Whatever you imagine as the worst,
multiply it by ten and you might get there. Today I was tasked to photo
rig the ship and surrounding area. It looked so much worse than I had
imagined, unbelievable really, with debris and disarray everywhere, the
ship listing, the hole in her side. I wish I had the power to relay to you
all what I have seen, but words just won't do it. I do want to tell you
the first thing that jumped out at me - the Stars and Stripes flying. I
can't tell you how that made me feel...even in this God forsaken hell hole
our flag was more beautiful than words can describe. Then I started to
notice the mass of activity going on below, scores of people working
non-stop in 90 plus degree weather to save this ship. They're doing it
with almost no electrical power and they're sleeping (when they can sleep)
outside on the decks because they can't stand the smell or the heat or the
darkness inside. They only want to eat what we bring them because they're
all scared of eating something brought by the local vendors. Even with all
that, the USS COLE and her crew is sending a message guys, and it's that
even acts of cowardice and hate can do nothing to the spirit and pride of
the United States. I have never been so proud of what I do, or of the men
and women that I serve with as I was today. There are sixteen confirmed
dead sailors who put it on the line for all of us, and some of them are
still trapped here. Please take a minute to pray for their families and
say a word of thanks for their sacrifice - one made so that we can live
the lives that we do. All of you that serve with me, thank you. All of you
that have loved ones that serve, thank you."
Please feel free to pass this on to those you think
will appreciate it.
Subj: Power cycling
It's not just ATX power supplies...
One of the first things I learned when learning to
fly the F-15E was that about 95% of all system glitches (radar,
navigation, displays, jammers, etc) can be solved this way, progressing to
the next step until the system starts working again:
1) Perform built-in-test cycle and clear error codes
out of computer memory 2) Perform soft reset (on-standby-on, or reset
switch) 3) Perform hard reset (some systems have a hard-reset switch which
also dumps memory) 4) Perform power cycle on malfuctioning system 5) Turn
off malfuctioning system, go through steps 1-4 on display processor, turn
system back on 6) Turn display processor off, power cycle malfunctioning
system, turn display processor on 7) Turn off malfunctioning system, cycle
power on central computer (dumps all navigation and weapons data), run BIT
tests on all required systems, turn on malfunctioning system 8) Turn off
malfunctioning system, forget about it for a while, turn it back on
sometime later in the mission.
This isn't exactly a written procedure, but I would
generally follow this progression of steps starting when the system glitch
was noticed until engine shutdown. If the system still didn't work by the
time I landed, check the computer for error codes and turn the whole mess
over to maintenance. Minor glitches and failures were common, but rarely
would a system remain completely unusable.
Sean Long seanlong@micron.net
My experience is similar, and I tend to pull the
plug between resets, but I can remember old linear power supplies with
half a Farad of capacitance. On the old S-100 bus systems you could
pull the plug and shove it back in again fast and the machines wouldn't
know you had done that...
Hi Jerry,
Dave Farquhar makes a good point when he states:
"An ATX computer that's powered off is
never truly powered off unless it's unplugged--you can get an indication
of this by watching the Wake-on-LAN lights on certain network cards.
"
We should remember this when servicing the newer
machines. I know it had been my habit to not disconnect the power cable on
XT and AT boxes so that I could physically ground myself when changing
boards, or use a grounded wrist strap grounded to the case. Ground can be
difficult to find in a residential or office environment.
Now with the ATX machines, it's not a good idea to
do this as the motherboard is never truely powered down and boards will
still have some power applied even in the 'supposed' powered off state.
Consequently, I'm less likely to use good antistatic techniques, now. Not
good, but replacing boards with 'any' amount of power applied is surely
less safe.
John rice@vx5.com
TOP
|
CLICK HERE for
a Special Request. 
