[no subject]

-JMS

-----Original Message-----
From: Paul <paul hessels sheridanc on ca>
To: redhat-list redhat com <redhat-list redhat com>
Date: Tuesday, March 02, 1999 12:23 PM
Subject: Re: 10/100 NIC Recommendations?


>Auto-sensing is the most useless thing in the world.  I have dealt with
>dozen of manufacturers of nics and switches and I have yet to find _one_
>that could make it work 100% of the time.  I have had _so_ many problems
>with this, that this is the first thing we now check if there is any
>network problems.
>
>It is useless.  Don't trust it.... don't trust anyone:)
>
>Sam Bayne wrote:
>>
>> At 02:04 AM 3/2/99 +0000, you wrote:
>> >Tony Johnson wrote:
>> >
>> >>
>> >> > 10-base-T uses two pairs.  1,2 and 3,6.
>> >> > 100-base-TX uses the same.
>> >> > 100-base-T4 uses all four pairs(8 wires)
>> >> >
>> >
>> >But then I would think that my crosscable I wired, which works fine with
>> >10base, would also work with 100base.  However it does not.  I was
pretty
>> >careful to make sure the twisted pairs were still twisted as close to
the
>> >rj45 connector as possible.  I heard that makes a difference.
>> >
>> >This brings me back to a question I had tacked on to this originally,
can you
>> >run 100 base peer to peer or are you required to go through a hub?
>>
>> I have some empirical evidence that you MUST use a hub. Using Intel Ether
>> Express Pro 10/100 NICs, (only at 10Mbit) and the eepro100 modular
driver,
>> we were able to connect two hubs no sweat.  When we tried to use a
>> crossover cable, to another 10mbit workstation, or to a 10 mbit
>> Utp-to-Fiber transceiver we got nothing.  We have used crossovers
>> periodically from hub-to-hub (temporary solutions while awaiting fiber
>> transceivers, switch ports, etc) and also from workstation-to-workstation
>> (Quick drive imaging, testing routing rules, etc) so we have known-good
>> crossover cables, and people who know how to make them.  My belief was th
at
>> the problem WASN'T with the crossed-over 100mbit, but that the crossover
>> somehow breaks the cards ability to sense whether to go 10 or 100.
>>
>> I haven't bothered to look at standards documents, so again this is only
>> uncontrolled experimental data.
>> --------------------------------------------------
>> Sam Bayne       -NT,Novell,HP-UX,Linux Sysadmin
>> sbayne nsccux sccd ctc edu           (206) 527-3762
>> North Seattle Community College/Computing Services
>> 9600 College Way N/Seattle,WA 98133
>> --------------------------------------------------
>>
>> --
>>   PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST
ARCHIVES!
>>                 http://www.redhat.com http://archive.redhat.com
>>          To unsubscribe: mail redhat-list-request redhat com with
>>                        "unsubscribe" as the Subject.
>
>
>--
>  PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
> http://www.redhat.com http://archive.redhat.com
>         To unsubscribe: mail redhat-list-request redhat com with
>                       "unsubscribe" as the Subject.
>