[Tfug] Switch problem

[Bexley Hall]

Hi Erich,

On 3/18/2013 11:57 AM, erich wrote:
> So far so good,

I think the captain on the Titanic was overheard making a similar
comment "earlier that evening"  ;-)

> I've rearranged the two switches, and they are powered up 24-7, and
> no problems have showed up.

Note that you might stumble on problems when everything is powered
off and then back on, again.  "In theory", you shouldn't -- but
practice seems to contradict theory!  :-/  (at least IME)

> Thus far only one of my machines registers
> "full duplex" that is 2000Mbps. It's brand new. I have another machine
> that's relatively new, and I'm wondering why that's not also full duplex.
> I'm thinking that the intervening 50' cable might be de-rating it.

Swapping cables is a cheap test/fix (assuming you have a spare!  :>)

But, check to be sure the "another machine" is truly set to
autonegotiate.  If only one end of a link autonegotiates,
you can end up with a HDX link.  If the "other" end is
operating in FDX, you end up with a "duplex mismatch"...
which gives LESS bandwidth than a HDX link!  (but, only
when you "tax" the link -- i.e., intermittent traffic will
*appear* unaffected).

Can you query the switch *and* host to see what *each* end
of the link thinks is happening?  Alternatively, moving some
big files (each way) and *measuring* throughput... (or, if
you can access error counters and watch for runts, late
collisions, etc.)

[I realize this may be impractical with SOHO kit]