problems with network setup sis900 in 10baseT mode

Rob van Dam rvand000 at wanadoo.nl
Tue Jul 29 15:37:10 PDT 2003


When problems come up in the near future, I will be the first one to let you 
know. I will have a further look at your wireless solutions (looks very 
interesting).

I do agree with you that this is a bad idea when you use the network for 
business. For me it's only for "home use" and reliability is not a very big 
issue for me.



On Tuesday 29 July 2003 22:13, Dagmar d'Surreal wrote:
> On Tue, 2003-07-29 at 12:50, Rob van Dam wrote:
> > I finally got it working I used the following startup script:
> >
> > ifconfig eth0 up 10.0.0.159 netmask 255.255.255.0
> > sleep 5
> > mii-diag -A 10baseT-FD eth0
> > sleep 10
> > mii-diag -A 10baseT-FD eth0
> > sleep 5
> >
> > After the "random" script I start up dhcpcd.
> >
> > And yes everything works perfect!
>
> Doubt it.  Without the proper number of twists in the cable a nearby
> hair-dryer or car with a noisy electrical system (for two examples) is
> going to murder your throughput.  The fact that the network card and
> hub/switch aren't negotiating properly should be your first sign that
> there's some serious signal loss taking place.
>
> Note, I'm not trying to come down on you in particular, I just want to
> make sure the other people reading on the list understand that this is a
> +very+ bad thing to do and will always cause massive headaches.
>
> > >That's a shame.  Try something wireless.  Telephone cable is not
> > >ethernet.  I'd catch hell on this list if I told you _exactly_ what I
> > >thought of someone trying to run 10base-T over telephone cable...
> > >although in this case I suspect the vast majority of them are thinking
> > >the exact same thing.  If you want to use that wire, stick a 33.6k
> >
> > modem
> >
> > >on each end.
> >
> > Wireless modems have a reach up to 100 meters through free air (not
> > through walls etc.) So this is no option to me.
>
> They go through walls, just not as far.  Masses of people (who contain
> much water) do a much better job of interfering with the signal.  Search
> Google for "pringles can wireless" for a _cheap_ solution to extending
> the range of a wireless network.  (I frequent a coffee house that has
> wireless and we can use the AP there from the porch outside with no
> problems, and literally 150 yards away from the building with a pringles
> antenna.)  Seriously, a pringles antenna is so cheap if you can find two
> people with notebooks and 802.11 cards you can do a little
> experimentation (and probably scare some people who thought their AP was
> secure in the process).
>
> In any case, unless that under the street conduit was laid with no room
> to spare, any decent electrician could probably run a snake through it
> to pass a new cable across.  (Might be something to keep in mind for
> possible upgrades to 100mbit)
>
> > >> It works perfectly in Windows so why should it work under Linux?
> > >
> > >"It works perfectly in Windows" is a lame response and don't expect me
> > >to fall for it.  Generally the only reason people get this idea is
> > >because Windows doesn't let them notice how _badly_ things are
> >
> > performing.  >Additionally... I am _pretty damn certain_ that it is
> > *not*
> >
> > >working perfectly under Windows.
> >
> > Not for me. It was the proof that it must be possible to get everything
> > working, and a big motivation to keep on trying.
> >
> > And it works, believe me. (I spend hours on internet by now, downloading
> > lots of stuff for example a complete Mandrake distribution, without
> > problems)
>
> Just wait for your first electrical storm, construction on the street,
> or someone operating a power drill nearby.
>
> > >Cisco can get gigE over old coathangers, but you are clearly not them.
> > >If you want to run your network over piano strings and foil bubblegum
> > >wrappers, that's your decision, but don't expect other people to
> >
> > support
> >
> > >it.
> >
> > I am not an expert, but I read several sites explaining that a 10 mbit
> > network is possible through telephone cable, because a 10 mbit network
> > doesn't cause much interference etc. and 100 mbit does.
>
> You should read more carefully then.  100mbit does not _cause_
> interference--it's much more susceptible to interference because of the
> increased complexity of the signal involved.  10mbit will behave fine on
> an improperly wired cable (i.e., when the pairs aren't matched up so
> they don't twist around each other, as when people mistake something
> wired as an RS-232 extension for an ethernet cable) provided there is
> almost _no_ electrical noise nearby.  Put them anywhere near something
> that generates a little bit of RFI (like an extension cord, or gods
> forbid, a fluorescent light) and packets will simply cease to traverse
> the wire.  I've been tanked before by someone putting a radio alarm
> clock _near_ a miswired cable. [1]  Those sites should probably mention
> that it will only work with _some_ phone cables, as not all telephone
> cable has the pairs twisted, as it is not required.
>
> With a _properly wired_ cable you can hang the 10base-T segment along
> the same hooks as an active string of common christmas lights and it
> won't be bothered (done that, too.  heheh).
>
> [1] Sidebar... Those cool looking "plasma balls" one finds for sale in
> gift shops generate enough noise to stop half a rackful of switches
> cold, proper cabling or not.  You don't want to know how I know this.
> --
> The email address above is just as phony as it looks, and for obvious
> reasons. Instant messaging contact nfo: AIM: evilDagmar  Jabber:
> evilDagmar at jabber.org




More information about the blfs-support mailing list