Bit of a mystery

[Michael C]

A customer tells me his entire network went down so that he could not access
any network drives on the server. I could not connect to the server via
remote desktop which is something that is always available to me. He
restarted the server but this didn't help. 20 minutes later he restarted the
receptionists machine at the front desk and everything went back to normal.
Is it even possible for a single machine to interupt the entire network? No
one would have done anything stupid like play with IP addresses or anything.
The front desk machine did crash and refused to restart and appeared to
interupt all network operation until the power plug was pulled from the
wall. This machine does appear to be a little unstable in that they have
reported it turning off without warning before (suddenly just goes to black
screen but fans continue torun). Any ideas on what could have happened?

Thanks in advance,
Michael

 

[GlowingBlueMist]

A customer tells me his entire network went down so that he could not
access any network drives on the server. I could not connect to the server
via remote desktop which is something that is always available to me. He
restarted the server but this didn't help. 20 minutes later he restarted
the receptionists machine at the front desk and everything went back to
normal. Is it even possible for a single machine to interupt the entire
network? No one would have done anything stupid like play with IP
addresses or anything. The front desk machine did crash and refused to
restart and appeared to interupt all network operation until the power
plug was pulled from the wall. This machine does appear to be a little
unstable in that they have reported it turning off without warning before
(suddenly just goes to black screen but fans continue torun). Any ideas on
what could have happened?

Thanks in advance,
Michael

Yes it is possible for an individual pc to bring down a network.

The PC could also have malfunctioned and seized the same IP address as the
server. Especially if the rest of the network could still access other
devices or the internet while the problem was occurring.

The offending PC may have been streaming constant garbage into the Ethernet.
Some routers have the ability to sense this kind of streaming garbage and
cut off the data from the offending port but they are not too common. If
they are using what is commonly known as a "dumb" bridge then nothing could
have prevented this kind of problem.

It's also possible that the Ethernet card or power supply is defective
causing abnormal voltages over the Ethernet cable to the bridge/router
causing a lock up. Might also explain the other problems the PC is/was
having with abnormal shutdowns. Depending on what is bad with a power
supply the fans may continue to run while the PC and other things fail.

Bad power supply, Ethernet card, or a partially corrupt operating system...
At this time it's anyone's guess.

 

[DanielEKFA]

Yes it is possible for an individual pc to bring down a network.

The PC could also have malfunctioned and seized the same IP address as the
server. Especially if the rest of the network could still access other
devices or the internet while the problem was occurring.

The offending PC may have been streaming constant garbage into the
Ethernet. Some routers have the ability to sense this kind of streaming
garbage and cut off the data from the offending port but they are not too
common. If they are using what is commonly known as a "dumb" bridge then
nothing could have prevented this kind of problem.

It's also possible that the Ethernet card or power supply is defective
causing abnormal voltages over the Ethernet cable to the bridge/router
causing a lock up. Might also explain the other problems the PC is/was
having with abnormal shutdowns. Depending on what is bad with a power
supply the fans may continue to run while the PC and other things fail.

Bad power supply, Ethernet card, or a partially corrupt operating
system... At this time it's anyone's guess.

I'd like to add to this, that if it were indeed some malfunctioning
hardware, causing not only electrical trouble on the ethernet, but also the
crashing of the machine, make sure your PCs are grounded! This is actually
not unimportant - we're used to seeing "This Apparatus Must Be Grounded" on
the back of much of our equipment, like amplifiers and DVD players, and
ignoring it safely (many times the power connector doesn't even have a
ground). PCs can build up massive amounts of static electricity if not
grounded. I obviously wouldn't recommend anyone to try this, but you can
actually kill yourself pretty easily by surfing for a couple of hours on an
un-grounded PC, then grabbing a radiator and the PC's casing, thus grounding
the system with your body. I repeat: don't try this!

Anyway, this hopefully won't be the case, since just one node in a connected
network needs to be grounded. I can't think of a single connector (save for
fibre networking, or wireless connectors obviously), which doesn't carry a
ground, so if the server (or one of its clients) is grounded, then all nodes
on its (wired) network are too.

 

[Rod Speed]

I'd like to add to this, that if it were indeed some malfunctioning
hardware, causing not only electrical trouble on the ethernet, but
also the crashing of the machine, make sure your PCs are grounded!

It wont have been that, because rebooting the receptionist's PC wouldnt have fixed that.

This is actually not unimportant - we're used to seeing "This
Apparatus Must Be Grounded" on the back of much of our equipment,
like amplifiers and DVD players, and ignoring it safely (many times
the power connector doesn't even have a ground). PCs can build up
massive amounts of static electricity if not grounded.
Nope.

I obviously wouldn't recommend anyone to try this, but you can actually kill yourself pretty easily by surfing for a
couple of hours on an un-grounded PC, then grabbing a radiator and the PC's casing, thus grounding the system with
your body.
Nope.

I repeat: don't try this!

Plenty do that in the millions of PC users using PCs and we
dont see any reports of anyone ending up dead that way.

Anyway, this hopefully won't be the case, since just one node in a connected network needs to be grounded.

You've mangled that completely too.

I can't think of a single connector (save for fibre networking, or wireless connectors obviously), which doesn't carry
a ground,

CAT5 doesnt.

so if the server (or one of its clients) is grounded, then all nodes on its (wired) network are too.

Wrong.

 

[Michael C]

The PC could also have malfunctioned and seized the same IP address as the
server. Especially if the rest of the network could still access other
devices or the internet while the problem was occurring.

The offending PC may have been streaming constant garbage into the
Ethernet. Some routers have the ability to sense this kind of streaming
garbage and cut off the data from the offending port but they are not too
common. If they are using what is commonly known as a "dumb" bridge then
nothing could have prevented this kind of problem.

It's also possible that the Ethernet card or power supply is defective
causing abnormal voltages over the Ethernet cable to the bridge/router
causing a lock up. Might also explain the other problems the PC is/was
having with abnormal shutdowns. Depending on what is bad with a power
supply the fans may continue to run while the PC and other things fail.

Bad power supply, Ethernet card, or a partially corrupt operating
system... At this time it's anyone's guess.

Thanks, that sounds like a good explanation. I can start with that as an
assumption and see if I can solve the problem from there. The system is
running a program called betwin which enables the system to run as if it's 2
computers with 2 screens, keyboards and mice. I suspect that is causing
problems so I think I'll put in an actual second PC and see if that solves
the problem.

Thanks for the reply.

Michael

 

[Rod Speed]

So Rod, which planet are you from anyway?

Same one you are, unfortunately.

On earth we seldom if ever find people surfing the net
while grabbing both their system case and a radiator.

A radiator aint the only grounded object around, cretin.

These three acts tend to be separated for some odd
reason, if one is feeling compelled to grab a radiator at all.

Wota terminal ****wit.

I hate to agree

No surprises there...

but you have a point here,

I did with the other point too.

and in fact I've seen more surge damage from switches or
hubs that weren't earth grounded, their PSUs weren't plugged
into a surge protector - then that surge traveled down the
ethernet cable seeking ground through the host computers.

Separate matter entirely to the stupid claim that surfing for a couple
of hours has any effect what so ever on the static buildup with a non
grounded PC, let alone whether static build up with a non grounded
PC can actually kill you. Of course it cant, even if you do ground
yourself to something grounded when using the PC.

 

[DanielEKFA]

Ignoring your rudeness, would you care to elaborate on all the "nope"s and
"totally mangled that"s. Some of us are here to learn, not to bash, so I for
one would like to know where and why I'm wrong, especially about why a
grounded PC wouldn't share its ground with other PCs on a network, when
ground is carried through ethernet cables?

TIA,
Daniel

 

[Michael C]

Ignoring your rudeness, would you care to elaborate on all the "nope"s and
"totally mangled that"s. Some of us are here to learn, not to bash, so I
for one would like to know where and why I'm wrong, especially about why a
grounded PC wouldn't share its ground with other PCs on a network, when
ground is carried through ethernet cables?

I'm guessing the cards themselves don't attach the gnd wires in the ethernet
cable to the gnd of the PC.

Michael

 

[Joel]

Ignoring your rudeness, would you care to elaborate on all the "nope"s and
"totally mangled that"s. Some of us are here to learn, not to bash, so I for
one would like to know where and why I'm wrong, especially about why a
grounded PC wouldn't share its ground with other PCs on a network, when
ground is carried through ethernet cables?

TIA,
Daniel

It's very interesting, I would like to know where you get the idea, and
how you think the hole thing works.

I didn't read the whole thread to know what's going on or this is the very
first Mystery I ever read, and just base on this single message without the
quote of the original message (I already deleted them without reading), I
can see you may be little more than just learning.

So, give us small detail of what in your mind, and see if we can come up
with thing to agree or disagree with your thinking.

 

[GlowingBlueMist]

Thanks, that sounds like a good explanation. I can start with that as an
assumption and see if I can solve the problem from there. The system is
running a program called betwin which enables the system to run as if it's
2 computers with 2 screens, keyboards and mice. I suspect that is causing
problems so I think I'll put in an actual second PC and see if that solves
the problem.

Thanks for the reply.

Michael

Ah yes, it can always be interesting while working with a device (or person
for that matter) that has a multiple personality...

I also agree with DanielIEKFA and others about checking the electrical
outlet for a possible grounding issue as well. In the USA it's quite easy
using one of those three light outlet testers. A person simply plugs it
into the outlet and if the correct two lights out of the possible three come
on the outlet wiring is correct. Any other pattern of lights indicates a
problem which the the attached fault chart should identify. You can get
them for around $8.00 in Home Depot or Lowes among other places. The one I
bought a month ago also included a button to test Ground Fault Interrupter
(GFI) outlets or breakers.

I have found many outlets that were wired incorrectly and a few commercially
produced outlet strips that also were incorrectly wired. Yes a PC may work
while using a incorrectly wired outlet but problems usually turn up sooner
or later, including possible physical harm to the user.

 

[Rod Speed]

Ignoring your rudeness, would you care to elaborate on all the "nope"s and "totally mangled that"s.

Already did.

Some of us are here to learn, not to bash, so I for one would like to know where and why I'm wrong,

The most obvious problem with your claims is that you cant get
killed by static buildup, even if the PC isnt grounded and you
touch the PC case and something that is grounded. The most
you can get in that situation is a static discharge which wont
kill you even if you have a heart problem.

And the amount of time spent browsing has no effect on static buildup either.

The primary source of static buildup with a PC is plastic carpet and a very dry
environment, usually air conditioning in winter without adequate humidity control.

especially about why a grounded PC wouldn't share its ground with
other PCs on a network, when ground is carried through ethernet cables?

It isnt.

 

[Paul]

Ignoring your rudeness, would you care to elaborate on all the "nope"s
and "totally mangled that"s. Some of us are here to learn, not to bash,
so I for one would like to know where and why I'm wrong, especially
about why a grounded PC wouldn't share its ground with other PCs on a
network, when ground is carried through ethernet cables?

TIA,
Daniel

Ethernet is not grounded. Each Ethernet interface is transformer
isolated. That is why a bunch of computers can connect to another
networking device, without a grounding problem.

See Figure 9 on PDF page 35. The ground on the RJ-45 shell, is to
chassis ground. It is not electrically connected to the ethernet
wires themselves. The chassis ground is for dumping ESD. There
are capacitors connected to ground, but that is for AC grounding.
The idea is, there is no DC path to ground, and thus no ground
loops. (Chip drives the left hand side of the diagram, and the
RJ-45 on the right, is the connector on the back of your computer.)

http://download.intel.com/design/network/applnots/82562ET_AP414.pdf

I've probed the wires on the CAT-5 cable in the lab, with a scope,
and there are 10's of volts of noise imposed on the wires. But,
because the wires are arranged in differential pairs, and treated
differentially, the common mode noise (same level of noise on
each wire), subtracts at the receiver. The receiver sees a clean
signal and virtually none of the noise. That is why it works.

Note that Figure 9, with the "Bob Smith" termination plane, is
for 10/100BT Ethernet. 1000BT (Gigabit) Ethernet, uses all four
pairs of Ethernet wires. There are more transformer coils involved
in that one, but the principles with respect to no DC path to
ground, remain. There are also RJ-45 connector stacks, where
the transformers are integrated into the stack, so you don't see
a black plastic rectangle on the motherboard.

http://ww2.pulseeng.com/products/datasheets/HC500.pdf

HTH,
Paul

 

[w_tom]

I'm guessing the cards themselves don't attach the gnd wires in the ethernet
cable to the gnd of the PC.

As Paul noted, Ethernet (all kinds) are galvanically isolated from
the cards. All that talk about safety grounds would (should) be
completely irrelevant to that problem.

Others have proposed solutions. Those and about about 100 other
solutions also could explain what had happened. Only way to obtain a
useful answer here is to provide more (better) information. For
example, before fixing anything, did anyone record what every light on
every box was reporting (especially on the router)? Did anyone first
run some basic diagnostics (ie Ping) to first learn what really
existed? What are the IP addresses of the various units - especially
the router?

Yes, computers must be properly earthed so that failures (ie
galvanic isolation failure) doe not create a problem. So yes, if a
failure exists, you may have multiple problems. Too many want to only
fix one, see the system working, and therefore ignore those other
problems.

Does static buildup exist or cause problms? No and no if (and
again) because of the many other solutons and backup solutions that
make static irrelevant. If static creates a problem, does grounding
eliminate it? Maybe or not necessarily. But again, only useful
answer needs far more information than what was posted.

One computer is unstable. So again, don't fix it. First identify
the problem. Collect facts so that the problem is identified or so
that others with better knowledge might post something useful here.
What did its system (event) logs and Device Manager report? What
happened when comprehensive hardware diagnostics (that execute without
any operating system) were executed (which do not exist if computer is
from a less responsible manufacturer)?

The point: if anyone has a 'it might be this' answer, then the
answer is wild speculation; woefully incomplete. At this point, the
useful answers says what information must be collected.

 

[DanielEKFA]

Ethernet is not grounded. Each Ethernet interface is transformer
isolated. That is why a bunch of computers can connect to another
networking device, without a grounding problem.

Thanks Did some browsing, too, and found out the same. Guess that's a
protective strategy of the network, in case some station being plugged in
would otherwise be able to suddenly discharge some electrostatic buildup and
possibly interrupt or damage something in the network?

See Figure 9 on PDF page 35. The ground on the RJ-45 shell, is to
chassis ground. It is not electrically connected to the ethernet
wires themselves. The chassis ground is for dumping ESD. There
are capacitors connected to ground, but that is for AC grounding.
The idea is, there is no DC path to ground, and thus no ground
loops. (Chip drives the left hand side of the diagram, and the
RJ-45 on the right, is the connector on the back of your computer.)

http://download.intel.com/design/network/applnots/82562ET_AP414.pdf

I've probed the wires on the CAT-5 cable in the lab, with a scope,
and there are 10's of volts of noise imposed on the wires. But,
because the wires are arranged in differential pairs, and treated
differentially, the common mode noise (same level of noise on
each wire), subtracts at the receiver. The receiver sees a clean
signal and virtually none of the noise. That is why it works.

Basically like the way balanced audio works, right? Twice the signal
strength, virtually none of the noise?

Recently I've been wondering how the phone works without noising up though,
after moving an outlet in my girlfriends new apartment and discovering only
two wires being connected. I would've thought it would be like a balanced
microphone connection where you have hot, cold, and ground connectors. I
mean, can you really do balanced audio without a ground? Or is it maybe that
it's only the last few meters from some larger station somewhere in the
building that isn't balanced and the signal strength high enough to not
introduce audible noise?

Note that Figure 9, with the "Bob Smith" termination plane, is
for 10/100BT Ethernet. 1000BT (Gigabit) Ethernet, uses all four
pairs of Ethernet wires. There are more transformer coils involved
in that one, but the principles with respect to no DC path to
ground, remain. There are also RJ-45 connector stacks, where
the transformers are integrated into the stack, so you don't see
a black plastic rectangle on the motherboard.

http://ww2.pulseeng.com/products/datasheets/HC500.pdf

HTH,

Very much, thank you I sort of had the misconception that any signal
would need to be grounded for it to be readable (like I still believe is the
case for balanced audio, where you need the ground to "know what's the 0" so
to speak). And I've seen at least one PSU when taken apart that had a ground
wire directly attach to the metal casing. Add to that the fact that many of
the connectors from the expansion cards have their ground directly touching
their metal backplate which in turn touches the case which in turn touches
ground on the power cord, I kinda saw the whole deal as one big "we gotta
share the ground" system. Which I guess is still the case "within" the
computer system itself? I understand digital communications don't need to be
grounded now.

I remember not having my computer grounded previously, and getting some
nasting shocks when accidentally touching e.g. audio or composite video
connectors when hooking up the puter to the rest of my systems. After
grounding the PC, I can touch all these connectors with no "side-effects" on
my body

Cheers for the lesson, it was really helpful

Daniel