suddenly slow network
- bill -
(load visionfs profedit in about 5 seconds to a win98 box). _Something_
happened one weekend (no electrical storms!) and that monday:
the answering machine had fallen off it's shelf
two ports on the 3com 10/100 hub were dead
a bunch of simple enunciators that use SCR's and LEDs were on even
though they were off on Friday.
Ever since then the network has been dead slow( takes 30 seconds to
invoke visionfs profedit). The telltale on the hub flashes the 100 Mb
light but instead of flashing rapidly it flashes intermittently.
I plan to swap the hub this week, but I am sure that you folk can
suggest a more intelligent way of diagnosing the problem.
In addition to being a (continual) unix neophyte I am a network rank
amateur, having been weeded to the Host model until last year.
--
-bill-
Technical Service Systems - bi...@TechServSys.com
- bill -
flog me, beat me: I forgot to show the configuration:
SCO 5.0.5
patches: rs505a
oss471e
oss497c
0ss600a
reversion to the pre oss497 cron
Bill Vermillion
- bill - <bi...@TechServsys.com> wrote:
>I have a tcp network running at 100 Mbs. Until recently it ran quickly
>(load visionfs profedit in about 5 seconds to a win98 box). _Something_
>happened one weekend (no electrical storms!) and that monday:
> the answering machine had fallen off it's shelf
> two ports on the 3com 10/100 hub were dead
> a bunch of simple enunciators that use SCR's and LEDs were on even
>though they were off on Friday.
>Ever since then the network has been dead slow( takes 30 seconds to
>invoke visionfs profedit). The telltale on the hub flashes the 100 Mb
>light but instead of flashing rapidly it flashes intermittently.
So what caused the answering machine to fall off the shelf. That
may be part of the answer. Cleaning people? Someone plug in
something that overloaded equipment and blew a breaker, and turn on
surge hit something.
Why do you think it takes an electrical storm to torch equipment.
Lightning doesn't require a storm to happen - though that's it's
most common environment.
But something such as a car accident and something hitting and
knocking down a power pole in the immediate area could cause it.
In the not-immediate area this could also happen if a collision
caused a short and short circuited something. If a grid went off
line even in a neighboring county/state the switch over could
produce a surge. IOW if you are connected to a power-source there
are many things besides lightning that can do you in.
Local construction can do this also. I saw plastic encased
surge protecting power strips burning when an electrician who was
connecting a 440VAC air-conditioner accidentally connected that to
the 110V AC distribution system. Of course that is usually quite
apparent - but surges in neighboring systems or those connected to
the same distribution transformer though not directly connected to
the damaged area could also get hit.
The fact that you have dead ports on a hub indicates some
electrical anomly.
>I plan to swap the hub this week, but I am sure that you folk can
>suggest a more intelligent way of diagnosing the problem.
Before you swap the hub check the device connected to the dead
port. I've heard of cases where individual components were swapped
- the supposed dead component being swapped out - and the system
still didn't work - and the source was found to be dead. That was
swapped and it was still dead, and the first component was dead.
Until someone wised up they had replace all components individual
twice before realizing that whatever failed first took out the
connected component and then the bad one left in took out the new
replacement unit. Not a common problem but something to be aware
of.
Check EVERTHING before you connect them all an power up.
--
Bill Vermillion bv @ wjv.com
- bill -
I agree with you. All components on the network are surge and UPS
protected. That is why I thought of lightning. But you are right. We
lost a UPS when a local crane bridged the KV lines to the high voltage
distribution lines, but even that wasn't this widespread
>
> >I plan to swap the hub this week, but I am sure that you folk can
> >suggest a more intelligent way of diagnosing the problem.
I can use netstat to look at the statistics, but what would I expect if
the hub (or a nic) was issuing garbage ? The hub is a 10/100
autosensing hub so it is sort of a switch, but not really. I must
buffer each port before squirting it out to the others. I am not there
so I can not tell you the hub number, it is a 3com Smart something (not
a switch)
>
> Before you swap the hub check the device connected to the dead
> port. I've heard of cases where individual components were swapped
> - the supposed dead component being swapped out - and the system
> still didn't work - and the source was found to be dead. That was
> swapped and it was still dead, and the first component was dead.
I first swapped the 3com NIC in the server as that is what was connected
to the first dead port - still didn't run. so I moved a known good nic
to that hub port and discovered that the hub had two damaged ports
(adjacent).
>
> Until someone wised up they had replace all components individual
> twice before realizing that whatever failed first took out the
> connected component and then the bad one left in took out the new
> replacement unit. Not a common problem but something to be aware
> of.
Sure glad that didn't happen to me.
>
> Check EVERTHING before you connect them all an power up.
>
> --
> Bill Vermillion bv @ wjv.com
--
Bill Vermillion
- bill - <bi...@TechServsys.com> wrote:
>Bill Vermillion wrote:
>> In article <3982e491$0$48167$2c3e...@news.voyager.net>,
>> - bill - <bi...@TechServsys.com> wrote:
>> Why do you think it takes an electrical storm to torch equipment.
>> Lightning doesn't require a storm to happen - though that's it's
>> most common environment.
....
>> The fact that you have dead ports on a hub indicates some
>> electrical anomly.
>I agree with you. All components on the network are surge and UPS
>protected. That is why I thought of lightning. But you are right. We
>lost a UPS when a local crane bridged the KV lines to the high voltage
>distribution lines, but even that wasn't this widespread
By 'all components are surge protected' does that include all
connections besides the AC. Many people think that just using a
UPS will protect them, but if anything else is connected, printer,
modem, scanner, etc., and those are also not protected/isolated,
electrical anomolies can come in that way.
That measn you should have rs232 surge devices on any serial
connections, 4-wire supressors on any WAN connections, basically on
any interconnecting cables, except of course fibre.
>> >I plan to swap the hub this week, but I am sure that you folk can
>> >suggest a more intelligent way of diagnosing the problem.
>I can use netstat to look at the statistics, but what would I expect if
>the hub (or a nic) was issuing garbage ? The hub is a 10/100
>autosensing hub so it is sort of a switch, but not really. I must
>buffer each port before squirting it out to the others. I am not there
>so I can not tell you the hub number, it is a 3com Smart something (not
>a switch)
I can't help you on that one.
>> Before you swap the hub check the device connected to the dead
>> port. I've heard of cases where individual components were swapped
>> - the supposed dead component being swapped out - and the system
>> still didn't work - and the source was found to be dead. That was
>> swapped and it was still dead, and the first component was dead.
>I first swapped the 3com NIC in the server as that is what was
>connected to the first dead port - still didn't run. so I moved a
>known good nic to that hub port and discovered that the hub had two
>damaged ports (adjacent).
Well it definately sounds like some sort of power hit. Janitors
deny everything of course :-)
Lot's of luck.
Bill
tom w
foot) connection to earth ground? If not, then no effective
surge protection.
UPSes are not surge protectors. Surge protectors that die
from a surge are too undersized for effective protection.
Furthermore, plug-in UPSes connect direct to AC power when not
operating from a battery. They should be called battery
backup. They are not surge protectors. That internal UPS
switchover relay does not perform surge protection. No earth
ground means no effective surge protection. 50' of 14 AWG to a
fuse box, then 20 feet to the earth ground rod is, to lightning,
is no earth ground; too far away.
Most assume that surge protectors stop lightning. Surge
protectors shunt, short, or divert a surge from one wire to all
others. Surge protectors don't stop or absorb lightning.
Another critical component for a surge protection 'system' -
earth ground. The 'surge protector' component passes a surge to
the 'earth ground' system component. Missing components means
that the entire system is defective. Where are the dedicated
ground wires on your existing surge protectors and UPSes?
Plug-in surge protectors are too close to transistors, too far
from earth ground, too expensive to only protect one device, and
too undersized for effective protection. Surges on each wire
must be shunted to earth ground where the wire enters the
building. Once inside, lightning surges find many conductive
materials such as the linoleum tile and latex paint. Effective
protection must be at least 1000 joules, 50,000 amps and be
directly connected to the common earth ground network - to keep
every surge out of the building.
Telephone COs, with overhead wires exposed everywhere, work in
every lightning storm, without damage. The technology has been
used successfully since the 1930s. Unfortunately we still build
as if the transistor did not exist. **No earth ground means no
effective surge protection**. Your UPSes and surge protectors
did not discuss critical earth ground on their boxes? What did
they forget to tell you - and why?
- bill - <bi...@TechServsys.com> wrote:
>I agree with you. All components on the network are
>surge and UPS protected. That is why I thought of
>lightning. But you are right. We lost a UPS when a
>local crane bridged the KV lines to the high voltage
>distribution lines, but even that wasn't this widespread
-----------------------------------------------------------
Got questions? Get answers over the phone at Keen.com.
Up to 100 minutes free!
http://www.keen.com
Jeff Liebermann
wrote:
> the answering machine had fallen off it's shelf
> two ports on the 3com 10/100 hub were dead
> a bunch of simple enunciators that use SCR's and ...
EMP/T or HERF. These are normally associated with nuclear detonations
and electronic warefare. HERF (High Energy Radio Frequency) and EMP
(Electro-Magnetic Pulse/Transmitter) are designed specifically to destroy
or cripple electronics. Considerable government research has been done
on these devices. Non-nuclear devices can also be easily constructed by
clueful individuals. See:
"Information Warefare" by Winn Schwartau 2nd ed ISBN 1-56025-132-8.
Chapter 10 covers the topic. Also see:
http://dirt.dmoz.org/Computers/Hacking/HERF,_EMP,_Tempest/
If this system is located near a secret government or research
organization, I would be seriously worried. Also, if you have some
clever 15 year olds, building weird looking devices, I would be even more
worried.
If it's not an EMP or HERF attack, then perhaps alien visitors might be
at fault. Are you near Roswell NM or Area 51?
Drivel: I spent a few hours doing battle with a small network that was
working just fine, but then mysteriously started belching errors and
collisions. Nothing had obviously failed, it was just slow. The network
was in a print shop and the central Linksys 16port 10baseT hub was
accessible to customers (and their kids). Someone had switched the power
supply voltage switch on the hub from 117VAC to 220VAC. The hub still
had enough power in this configuration to run the lights, and move some
packets, but data errors were epidemic. I didn't notice the switch
problem until I had replaced the hub (fixing the problem) and was
dragging the carcass to my truck.
--
Jeff Liebermann je...@comix.santa-cruz.ca.us
150 Felker St #D Santa Cruz CA 95060
831-421-6491 pager 831-429-1240 fax
http://www.cruzio.com/~jeffl/sco/ SCO stuff
Bill Vermillion
tom w <bareface...@juno.com.invalid> wrote:
Just a comment on a couple of things.
> UPSes are not surge protectors. Surge protectors that die
>from a surge are too undersized for effective protection.
>Furthermore, plug-in UPSes connect direct to AC power when not
>operating from a battery. They should be called battery
>backup.
They give a false sense of security.
At one time the industry did differentiate from a true UPS and what
was then called a 'standby power supply'. Now everyting is called
UPS. BEST made some wonderful units with torroid transformers.
The system always ran off the battery and the battery was
continually charge from the mains. Priced at about three times
more than other devices inaccurately labeled UPS a lot of people
didn't buy them that really needed those devices.
> Another critical component for a surge protection 'system' -
>earth ground. The 'surge protector' component passes a surge to
>the 'earth ground' system component. Missing components means
>that the entire system is defective. Where are the dedicated
>ground wires on your existing surge protectors and UPSes?
This is not neccesarily 100% correct - at least here in lighting
coutnry. Cheap devices will have 'protection' on the two power
sides and shunt that to ground and don't even consider putting
anything on the ground side. In this area you need to make sure
that hot, neutral and ground are all protected as we get probably
just as many lighting problems with lightning entering through the
ground and the out through the power lines.
The worst any client of mine had was a ground strike between their
two buildings and went both directions and took out terminals,
PC's, terminal ports and printers in both building. That was about
$25K just on equipment replacement. You see some intersting
'ceramic looking' shapes from the melted sand where these hit.
I don't know where the strike came from at another customer site
two weeks ago. That was about $5K total including parts and labor
but excluding the T1 channel bank. That was probably another $2K
minimum. The channel bank was the first piece of telco equipment
I've seen torched in many years.
> Telephone COs, with overhead wires exposed everywhere, work in
>every lightning storm, without damage. The technology has been
>used successfully since the 1930s. Unfortunately we still build
>as if the transistor did not exist. **No earth ground means no
>effective surge protection**. Your UPSes and surge protectors
>did not discuss critical earth ground on their boxes? What did
>they forget to tell you - and why?
During one really nasty storm I was one of the few that had NO
problems - for just the same reasons you give - the house was old -
probably about 70 years - and it was two-wire only. No earth
ground at all anywhere. The whole building floated.
As to some current telco/carrier facilities - one I have some
co-located machines in has NO metal data connections into the
building. And the fiber enters/exits from three different locations.
>
>- bill - <bi...@TechServsys.com> wrote:
>>I agree with you. All components on the network are
>>surge and UPS protected. That is why I thought of
>>lightning. But you are right. We lost a UPS when a
>>local crane bridged the KV lines to the high voltage
>>distribution lines, but even that wasn't this widespread
>
>
>
>-----------------------------------------------------------
>
>Got questions? Get answers over the phone at Keen.com.
>Up to 100 minutes free!
>http://www.keen.com
>
Ben Rosenthal
Bill Vermillion wrote:
> In article <3982e491$0$48167$2c3e...@news.voyager.net>,
> - bill - <bi...@TechServsys.com> wrote:
>
> >I have a tcp network running at 100 Mbs. Until recently it ran quickly
> >(load visionfs profedit in about 5 seconds to a win98 box). _Something_
> >happened one weekend (no electrical storms!) and that monday:
> > the answering machine had fallen off it's shelf
> > two ports on the 3com 10/100 hub were dead
> > a bunch of simple enunciators that use SCR's and LEDs were on even
> >though they were off on Friday.
>
> >Ever since then the network has been dead slow( takes 30 seconds to
> >invoke visionfs profedit). The telltale on the hub flashes the 100 Mb
> >light but instead of flashing rapidly it flashes intermittently.
>
> So what caused the answering machine to fall off the shelf. That
> may be part of the answer. Cleaning people? Someone plug in
> something that overloaded equipment and blew a breaker, and turn on
> surge hit something.
>
I like this theory alot. Some one one was trying to do something with the
phone system or answering machine. They plugged telephone equipment into a
data port some where in the building or directly into the hub (the plugs will
fit). As I remember telephone equipment will fry network hardware (i know
RS-232 will).
<snip-o-rama>
> --
> Bill Vermillion bv @ wjv.com
Ben Rosenthal
- bill -
>
> In article <39842ece$0$48157$2c3e...@news.voyager.net>,
> - bill - <bi...@TechServsys.com> wrote:
> >Bill Vermillion wrote:
>
> >> In article <3982e491$0$48167$2c3e...@news.voyager.net>,
> >> - bill - <bi...@TechServsys.com> wrote:
>
> >> Lightning doesn't require a storm to happen - though that's it's
> >> most common environment.
> ....
> >> The fact that you have dead ports on a hub indicates some
> >> electrical anomly.
>
> >protected. That is why I thought of lightning. But you are right. We
> >lost a UPS when a local crane bridged the KV lines to the high voltage
> >distribution lines, but even that wasn't this widespread
>
> connections besides the AC. Many people think that just using a
> UPS will protect them, but if anything else is connected, printer,
> modem, scanner, etc., and those are also not protected/isolated,
> electrical anomolies can come in that way.
>
> That measn you should have rs232 surge devices on any serial
> connections, 4-wire supressors on any WAN connections, basically on
> any interconnecting cables, except of course fibre.
This is a really easy one to answer (which, of course means the answer
will be wrong) system is server + hub + 4 pcs and 1 printer. No
external connections. All but the printer are plugged into upss.
All are from the same power panel with isolated power to the server.
>
> >> >I plan to swap the hub this week, but I am sure that you folk can
> >> >suggest a more intelligent way of diagnosing the problem.
>
> >I can use netstat to look at the statistics, but what would I expect if
> >the hub (or a nic) was issuing garbage ? The hub is a 10/100
> >autosensing hub so it is sort of a switch, but not really. I must
> >buffer each port before squirting it out to the others. I am not there
> >so I can not tell you the hub number, it is a 3com Smart something (not
> >a switch)
>
> I can't help you on that one.
>
> >> Before you swap the hub check the device connected to the dead
> >> port. I've heard of cases where individual components were swapped
> >> - the supposed dead component being swapped out - and the system
> >> still didn't work - and the source was found to be dead. That was
> >> swapped and it was still dead, and the first component was dead.
>
> >I first swapped the 3com NIC in the server as that is what was
> >connected to the first dead port - still didn't run. so I moved a
> >known good nic to that hub port and discovered that the hub had two
> >damaged ports (adjacent).
>
> Well it definately sounds like some sort of power hit. Janitors
> deny everything of course :-)
"uh, you mean that I shouldn't have brought my brother's welder with me
and used it to fix my bycycle ?"
>
> Lot's of luck.
I hate computers.
- bill -
>
> So each surge protector and UPS has a direct (less than 10
> foot) connection to earth ground? If not, then no effective
> surge protection.
>
> from a surge are too undersized for effective protection.
> Furthermore, plug-in UPSes connect direct to AC power when not
> operating from a battery. They should be called battery
> switchover relay does not perform surge protection. No earth
> ground means no effective surge protection. 50' of 14 AWG to a
> fuse box, then 20 feet to the earth ground rod is, to lightning,
> is no earth ground; too far away.
>
> Most assume that surge protectors stop lightning. Surge
> protectors shunt, short, or divert a surge from one wire to all
> others. Surge protectors don't stop or absorb lightning.
>
> earth ground. The 'surge protector' component passes a surge to
> the 'earth ground' system component. Missing components means
> that the entire system is defective. Where are the dedicated
> ground wires on your existing surge protectors and UPSes?
>
> from earth ground, too expensive to only protect one device, and
> too undersized for effective protection. Surges on each wire
> must be shunted to earth ground where the wire enters the
> building. Once inside, lightning surges find many conductive
> materials such as the linoleum tile and latex paint. Effective
> protection must be at least 1000 joules, 50,000 amps and be
> directly connected to the common earth ground network - to keep
> every surge out of the building.
>
> every lightning storm, without damage. The technology has been
> used successfully since the 1930s. Unfortunately we still build
> as if the transistor did not exist. **No earth ground means no
> effective surge protection**. Your UPSes and surge protectors
> did not discuss critical earth ground on their boxes? What did
> they forget to tell you - and why?
Thank you. You are right the "ground" is the electrical system ground
wire. The distribution panel does have a good earth ground - 6 gage but
the pcs are 20-60 feet from the panel. Ok, I accept we could have
ground loops with a big surge.
>
> - bill - <bi...@TechServsys.com> wrote:
> >I agree with you. All components on the network are
> >surge and UPS protected. That is why I thought of
> >lightning. But you are right. We lost a UPS when a
> >local crane bridged the KV lines to the high voltage
> >distribution lines, but even that wasn't this widespread
>
> -----------------------------------------------------------
>
> Got questions? Get answers over the phone at Keen.com.
> Up to 100 minutes free!
> http://www.keen.com
--
- bill -
>
> On Sat, 29 Jul 2000 10:05:05 -0400, - bill - <bi...@TechServsys.com>
> wrote:
>
> > the answering machine had fallen off it's shelf
> > two ports on the 3com 10/100 hub were dead
>
> EMP/T or HERF. These are normally associated with nuclear detonations
> and electronic warefare. HERF (High Energy Radio Frequency) and EMP
> (Electro-Magnetic Pulse/Transmitter) are designed specifically to destroy
> or cripple electronics. Considerable government research has been done
> on these devices. Non-nuclear devices can also be easily constructed by
> clueful individuals. See:
> "Information Warefare" by Winn Schwartau 2nd ed ISBN 1-56025-132-8.
> Chapter 10 covers the topic. Also see:
> http://dirt.dmoz.org/Computers/Hacking/HERF,_EMP,_Tempest/
>
> If this system is located near a secret government or research
> organization, I would be seriously worried. Also, if you have some
> clever 15 year olds, building weird looking devices, I would be even more
> worried.
>
many legislators/governor's aides yelling over the phones at the same
time could do that ? <G>
> If it's not an EMP or HERF attack, then perhaps alien visitors might be
> at fault. Are you near Roswell NM or Area 51?
Now, an alient teenager might be the problem. The adult aliens we talk
with are too nice to do anyting like this. Although, to think of it,
Elvis was a bit upset that weekend.
>
> Drivel: I spent a few hours doing battle with a small network that was
> working just fine, but then mysteriously started belching errors and
> collisions. Nothing had obviously failed, it was just slow. The network
> was in a print shop and the central Linksys 16port 10baseT hub was
> accessible to customers (and their kids). Someone had switched the power
> supply voltage switch on the hub from 117VAC to 220VAC. The hub still
> had enough power in this configuration to run the lights, and move some
> packets, but data errors were epidemic. I didn't notice the switch
> problem until I had replaced the hub (fixing the problem) and was
> dragging the carcass to my truck.
>
> --
> Jeff Liebermann je...@comix.santa-cruz.ca.us
> 150 Felker St #D Santa Cruz CA 95060
> 831-421-6491 pager 831-429-1240 fax
> http://www.cruzio.com/~jeffl/sco/ SCO stuff
--
Bill Vermillion
>> In article <3982e491$0$48167$2c3e...@news.voyager.net>,
>> - bill - <bi...@TechServsys.com> wrote:
>> >I have a tcp network running at 100 Mbs. Until recently it ran quickly
>> >(load visionfs profedit in about 5 seconds to a win98 box). _Something_
>> >happened one weekend (no electrical storms!) and that monday:
>> > the answering machine had fallen off it's shelf
>> > two ports on the 3com 10/100 hub were dead
>> > a bunch of simple enunciators that use SCR's and LEDs were on even
>> >though they were off on Friday.
>> >Ever since then the network has been dead slow( takes 30 seconds to
>> >invoke visionfs profedit). The telltale on the hub flashes the 100 Mb
>> >light but instead of flashing rapidly it flashes intermittently.
>> So what caused the answering machine to fall off the shelf. That
>> may be part of the answer. Cleaning people? Someone plug in
>> something that overloaded equipment and blew a breaker, and turn on
>> surge hit something.
>I like this theory alot. Some one one was trying to do something
>with the phone system or answering machine. They plugged telephone
>equipment into a data port some where in the building or directly
>into the hub (the plugs will fit). As I remember telephone
>equipment will fry network hardware (i know RS-232 will).
You mean something like the 90V ring current for telephones going
into something designed for 5 volts? You can be unpleasantly
surprised if you happen to be holding a phone pair when the dial
current is sent - imagine what it can do to a defenseless chip. :-)
Bill
Ben Rosenthal
Bill Vermillion wrote:
The clue is the answering machine being on the floor. This points to some one
messing around with phone equip. and like you say 90v into a hub should do some
damage. If this happened its a wonder any of it works.
<asside>
There was a post on this group where a janitor plugged a high power device into a
wall outlet that was on a UPS. The result was that it overloaded the UPS and
brought down all the "protected" equipment.
<end asside>
Ben Rosenthal
Ben Rosenthal
- bill - wrote:
> Jeff Liebermann wrote:
> >
> > On Sat, 29 Jul 2000 10:05:05 -0400, - bill - <bi...@TechServsys.com>
> > wrote:
> >
> > > the answering machine had fallen off it's shelf
> > > two ports on the 3com 10/100 hub were dead
> > > a bunch of simple enunciators that use SCR's and ...
> >
> > EMP/T or HERF. These are normally associated with nuclear detonations
> > and electronic warefare. HERF (High Energy Radio Frequency) and EMP
> > (Electro-Magnetic Pulse/Transmitter) are designed specifically to destroy
> > or cripple electronics. Considerable government research has been done
> > on these devices. Non-nuclear devices can also be easily constructed by
> > clueful individuals. See:
> > "Information Warefare" by Winn Schwartau 2nd ed ISBN 1-56025-132-8.
> > Chapter 10 covers the topic. Also see:
> > http://dirt.dmoz.org/Computers/Hacking/HERF,_EMP,_Tempest/
> >
> > If this system is located near a secret government or research
> > organization, I would be seriously worried. Also, if you have some
> > clever 15 year olds, building weird looking devices, I would be even more
> > worried.
> >
> I am 2 blocks from the state capitol of Michigan. Do you think that too
> many legislators/governor's aides yelling over the phones at the same
> time could do that ? <G>
>
Just one politician during election season. Or the big 3 lobbying the
politician during election season is likely. <G>
>
> > If it's not an EMP or HERF attack, then perhaps alien visitors might be
> > at fault. Are you near Roswell NM or Area 51?
>
> Now, an alient teenager might be the problem. The adult aliens we talk
> with are too nice to do anyting like this. Although, to think of it,
> Elvis was a bit upset that weekend.
That was the night my UPSs and servers formed a crop circle in the server room.
8-)
>
> >
> > Drivel: I spent a few hours doing battle with a small network that was
> > working just fine, but then mysteriously started belching errors and
> > collisions. Nothing had obviously failed, it was just slow. The network
> > was in a print shop and the central Linksys 16port 10baseT hub was
> > accessible to customers (and their kids). Someone had switched the power
> > supply voltage switch on the hub from 117VAC to 220VAC. The hub still
> > had enough power in this configuration to run the lights, and move some
> > packets, but data errors were epidemic. I didn't notice the switch
> > problem until I had replaced the hub (fixing the problem) and was
> > dragging the carcass to my truck.
> >
> > --
> > Jeff Liebermann je...@comix.santa-cruz.ca.us
> > 150 Felker St #D Santa Cruz CA 95060
> > 831-421-6491 pager 831-429-1240 fax
> > http://www.cruzio.com/~jeffl/sco/ SCO stuff
>
> --
>
> -bill-
>
> Technical Service Systems - bi...@TechServSys.com
Ben Rosenthal
Jeff Liebermann
wrote:
>I am 2 blocks from the state capitol of Michigan. Do you think that too
>many legislators/governor's aides yelling over the phones at the same
>time could do that ? <G>
No, but hot air has been known to overheat equipment. Coupled with the
usual bull and baloney found during election year politics, I would
suspect that you hub may have simply rolled over in disgust.
>> If it's not an EMP or HERF attack, then perhaps alien visitors might be
>> at fault. Are you near Roswell NM or Area 51?
>
>Now, an alient teenager might be the problem. The adult aliens we talk
>with are too nice to do anyting like this.
The first step to solving a problem is to blame someone. However, it is
often difficult to find a suitable perpertrator or scape goat.
Therefore, I find alien visitors and teenagers to be convenient, as both
have similar logic patterns and are equally destructive.
As an added thought, I've seen one UPS that was responsible for blowing
up equipment instead of protecting it. Its "modified sine wave" was more
like a spike infested square wave with lots of ringing and overshoot.
Whenever the power would die, mysterious failures appear in the allegedly
protected equipment. The final and fatal clue was when it blew itself
up, complete with fire, smog and brimstone. I'll pretend not to recall
the manufacturer as only this one unit was defective.
Speaking of UPS's, I have a large number of APC 1400RH UPS's. APC is an
international company, and therefore does not label their lights and
switches in English. Instead, they have weird symbols that only an
expert in ancient Egyptian hieroglyphics can understand. The "test"
button was at least labeled, but there was no indication that it also
functioned as an "on" button. My initial introduction into what the
other front panel push button does was followed by having much of a radio
site loose power. Then, I figgered it out. The "1" on the test button
means on. The "0" on the other button means "off". Must have been
designed by an engineer or programmer. I now have decoder charts and
English translations attached to many of these UPS's.
Jeff Liebermann
wrote:
>You mean something like the 90V ring current for telephones going
>into something designed for 5 volts? You can be unpleasantly
>surprised if you happen to be holding a phone pair when the dial
>current is sent - imagine what it can do to a defenseless chip. :-)
I've actually done that (I was bored). I shoved the ring generator
output from a KSU into an ethernet hub on both the tx and rx wires. No
fireworks, no failures. TIA 802.3 specifies in Section 14.7.2.5 that
ethernet is suppose to survive both the battery voltage (56VDC) and the
ringing voltage (175VAC P-P from 20 to 60Hz). It does because the input
transformer has almost no frequency response below 1Mhz. I won't claim
that all hubs can survive such abuse, but the one's I tried did.
- bill -
>
> I have a tcp network running at 100 Mbs. Until recently it ran quickly
> (load visionfs profedit in about 5 seconds to a win98 box). _Something_
> happened one weekend (no electrical storms!) and that monday:
> two ports on the 3com 10/100 hub were dead
> though they were off on Friday.
>
> Ever since then the network has been dead slow( takes 30 seconds to
> invoke visionfs profedit). The telltale on the hub flashes the 100 Mb
> light but instead of flashing rapidly it flashes intermittently.
>
> suggest a more intelligent way of diagnosing the problem.
>
> amateur, having been weeded to the Host model until last year.
>
>
> -bill-
>
> Technical Service Systems - bi...@TechServSys.com
I replaced the 3com 10/100 hub (dual speed hub 500) with a Linksys
10/100 hub. What a surprise. The linksys shows continual 100 Mb
activity on the line to the server NIC (rapid flashing on that link
light), but it doesn't seem to be going anywhere (none of the other link
lighs blink). netstat shows that I am getting 2:1 errors to good
packets. (BTW: with the linksys I can not ping the server from a win98
pc on the network) I replaced the Linksys with the 3com and, although
slow the network works. So I guess the next step is to replace the
server NIC 3com 3c905b as it seems to be spewing garbage.
Anyone have any thoughts on this diagnosis ?
Part II: I have never replaced a NIC in a 505 system yet. Can I use
the gui and just delete the NIC (and, I presume the associated protocol
and NFS runtime) and then swap cards and reinstall the nic/protocol/nfs
runtime) or is there an easier way ?
slowtolearn
Assuming it's the NIC (which the collision rate/constant activity
COULD indicate), if you get yourself another 3c905b, you should be
able to simply swap the cards and be on your way. I have done this in
5.0.4 successfully, I'm guessing in 5.0.5 it would be OK as well (?)
You wouldn't want to change any other cards in the box, as this will
mess with the PCI IRQ mapping, and you _may_ not get the original IRQ
on the new card.
If you need to, you can use the GUI, or scoadmin-> networks-> network
configuration manager, or netconfig to remove/add the card.
Scott
On Wed, 02 Aug 2000 09:30:48 -0400, - bill - <bi...@TechServsys.com>
wrote:
>- bill - wrote:
Jeff Liebermann
wrote:
I guess you've eliminated aliens and teenagers as possible culprits.
>(BTW: with the linksys I can not ping the server from a win98
>pc on the network) I replaced the Linksys with the 3com and, although
>slow the network works. So I guess the next step is to replace the
>server NIC 3com 3c905b as it seems to be spewing garbage.
Yep. Actually, most hubs will NOT show traffic from an insane NIC. Nor
will the traffic monitors that expect valid data instead of garbage.
>Anyone have any thoughts on this diagnosis ?
Yeah. What's happening is the the NIC receiver is not decoding MFM data
properly. It thinks it's getting "malformed" packets and is firing off
the "trashing" mechanism, by which ethernet intentionally trashes mangled
packets. The dual speed hub (I hate those) will not propogate collisions
and garbage across the internal bridge between the 100 and 10 side of the
hub. Therefore, your 10baseT devices probably work just fine, while your
100baseT machines may be having a difficult time.
>Part II: I have never replaced a NIC in a 505 system yet. Can I use
>the gui and just delete the NIC (and, I presume the associated protocol
>and NFS runtime) and then swap cards and reinstall the nic/protocol/nfs
>runtime) or is there an easier way ?
If it's another 3c905[a|b|c], then just plug it in and it's done. No
configuration. If it's a different flavour, you should probably backup
the major config files, remove the driver, shove in the new card, install
the driver, and you might luck out that the stack recognizes "net0" as
being the same as before and without reconfiguration. I did this once
and vaguely recall that all the configuration files were retained as long
as the driver was called net0. However, I was running in fire-drill
mode, and didn't have time to be sure.
tom w
>I am 2 blocks from the state capitol of Michigan. Do
>you think that too many legislators/governor's aides
>yelling over the phones at the same time could do that ? <G>
According the basic political theory, politicans cannot create
surges - only noise. However certain bills act as lightning
rods. The surge arrives simultaneously on airplanes from every
direction. They are called "lobbyists".
How do you build surge protection when electrons are replaced
by $100 bills?
- bill -
>
> - bill - <bi...@TechServsys.com> wrote:
> >Jeff Liebermann wrote:
> >I am 2 blocks from the state capitol of Michigan. Do
> >you think that too many legislators/governor's aides
> >yelling over the phones at the same time could do that ? <G>
>
> According the basic political theory, politicans cannot create
> surges - only noise. However certain bills act as lightning
> rods. The surge arrives simultaneously on airplanes from every
> direction. They are called "lobbyists".
>
> How do you build surge protection when electrons are replaced
> by $100 bills?
>
I just pipe them to my bank account.