[Wlug] ISP/Router/Modem Ethernet communications

[kstratton at fastmail.us]

Thank you everybody for your replies.  I now know that this
autonegotiation issue can be potentially be a much bigger problem with
100 base T than I thought.  Specifically, there is no defined mechanism
in the spec to autonegotiate for 100 base T cards.  If I actually have
time for this, I guess I will have to find the spec and read it...

My network driver (windows) is a "NVIDIA nForce MCP Networking
Controller".  It is an integrated network connector on a HP pavilion.  I
do not know of a windows equivalent to lspci, so I cannot get the actual
reported hardware rev.  Maybe I have to get a good live CD distro.

In this case, resetting the cable modem did not fix the problem,
although it may have fixed it in the past.

My router is a four port Linksys Router about 5 to 8 years old.  My
modem is a Speedstream 5280.  I am sufficiently annoyed by this problem
to consider buying a new router if this router has a force 10 base T
option for the WAN.  I am willing to listen to any suggestions.

  

On Fri, 30 Nov 2007 16:29:17 -0500, "John Stoffel" <john at stoffel.org>
said:
> 
> kstratton> I just went through some painful debug session(s) with my
> kstratton> ISP.  For some reason, after forcing 10 base T half duplex
> kstratton> through my windows box directly to my modem (I do not
> kstratton> expect ISPs to support linux),everything suddenly worked
> kstratton> (after reconnection and re-enabling of course).
> 
> Can you give more details on the cable modem box you're using and the
> PC as well?  Such as what network card in the PC you are using? 
> 
> kstratton> How often does this kind of thing happen?  I have seen this
> kstratton> kind of thing before only once before, and I was using an
> kstratton> old hub that only supported 10 base T, not a home router
> kstratton> that is supposed to autodetect the port type.  I remember
> kstratton> that the modem only supports 10 base T, but I am not 100%
> kstratton> certain.
> 
> Sun and Cisco were notirious for having problems figuring our
> autonegotiation on the Sun hme (Happy Meal Ethernet) cards.  They'd
> end up 100Full on one side and 100 Half Duplex on the other.  It would
> work, just very slowly...
> 
> kstratton> Does anybody have an explanation of what most likely
> kstratton> happened?  Do not hesitate to skimp on technical details or
> kstratton> references if is convenient.  I desperately want to
> kstratton> understand what happened.  --
>   
> Something didn't autonegotiate properly.  Did you try powering off
> both devices and then powering them on starting with the cable box?
> 
> Also, alot of Cable Companies lock the cable box to the first MAC
> (ethernet hardware address) they see coming over the link.  So if you
> boot up with a PC, and then try to move to something else like a NAT
> box or a linux box acting as a NAT box, things can go wonky.  
> 
> Usually a hard reset of the Cablemodem will do the trick, but
> sometimes you need to contact the ISP and ask them to reset it for
> you.
> 
> In your case, it really sounds like a problem with Autonegotiation.
> Some devices just don't do a good job, esp older equipment from before
> the Gigabit ethernet stage.  
> 
> Just to expand on Frank's email, the people who designed the Gigiabit
> Ethernet standard over Copper (802.xxx I can't remember) saw all the
> problems with the 100/10 devices and autonegotiation and explicity
> made AutoNeg part of the spec in a very well detailed way, so that
> these problems won't happen again.  As a matter of fact, Gigabit ports
> are much smarter and don't require crossover cables either, you can
> just plug them into back to back, they figure out what's going on
> automatically and adjust.  
> 
> The joys of big ASICs!  
> 
> John
> _______________________________________________
> Wlug mailing list
> Wlug at mail.wlug.org
> http://mail.wlug.org/mailman/listinfo/wlug
-- 
  
  kstratton at fastmail.us