[Tfug] Switch problem

Wed Mar 13 22:09:48 MST 2013

It is worth noting that the auto-negotiation signaling uses a VERY
different protocol and line frequencies than the subsequent data
transmission.  My guess is two GigaBit interfaces will auto-negotiate to 1
GigaBit on a physical connection that will only support 10MegaBit.  I
haven't seen that one, bit I have seen an auto-negotiation to 100MegaBit on
a line 600ft link that worked fine when forced to 10MegaBit.  The topology
I used for a 1200ft link was a 10MegaBit (custom POE) hub in the halfway
point.  It has worked fine for years (except when the hub got wet <sigh>).

On Wed, Mar 13, 2013 at 8:27 AM, erich <erich1 at copper.net> wrote:

> Yes,
>         I've isolated my switch. I have it mediating between two boxen at
> a time and discovered some interesting properties
>
>           1)  The little switch will tell you the capabilities of your
> box's
> interface card, 10/20Mbps, 100/200Mbps, and 2000Mbps
>
>           2) The problem shows up if everything is unplugged from
> power lines. In other words we're not even using stand-by pwer
> supplies, and the switch power supply was unplugged, (This
> is unusual. Network gear usually runs 24-7)
>
>           3) Old, slow boxen have the most trouble: It's not
> enough to reboot this old slow box of mine. I had to
> shut it down, Power it off, and bring it back up again
> before it finally started working networkwise.
>
>           4) If the switch is not shut down. Other boxen will
> recover and work normally when powered up with
> their network interfaces. (I hope this is true. I've not
> tested this proposition with my entire network). The
> little switch uses 3.5 watts. So leaving it run 24-7
> won't brake ta bank.
>
> Erich
>
>
>
> Bexley Hall wrote:
>
>> Hi Erich,
>>
>> On 3/11/2013 10:51 PM, erich wrote:
>>
>>> I bought
>>>
>>> a Trendnet TEG-S80g switch, and installed it in my network. I pro-
>>> ceded to be amazed at the speedup.
>>>
>>
>> <grin>  There was a time when "velocitized" only applied to driving!
>>
>>  I'm just a home user, and turn
>>> everything off at nite including the switch.
>>>
>>
>> I leave my fabric powered up as I never know <when> I'm going to
>> want to use <what>.
>>
>>  The next day troubles started when I powered up my system.
>>> One of my boxes waited for a DHCP address that never showed
>>> up. Tried pinging other boxen with disturbing "address unreachable"
>>> messages. It was a struggle to preserve my sanity. For a while
>>> it seemed that the switch had damaged the network interface of
>>> some of my PC's. I took the new switch out of the system.
>>>
>>> After rebooting several machines, the system started coming
>>> back up. "Damaged" network interfaces started working again.
>>>
>>
>> To be clear:  they are working with some *other* (non-Gb) switch (?)
>>
>>  This is a high-speed switch. Is it possible that It could interact
>>> with a network interface on a box and pull it into a state that
>>> would remain on poweroff?
>>>
>>
>> Gb interfaces have to autonegotiate lots of configuration parameters.
>> Among them, speed (10/100/1000), Full Dupled/Half Duplex, and usually
>> AutoMDIX.
>>
>> I've encountered SOHO switches that wouldn't talk to other *identical*
>> (same manufacturer/model number) switches in the past because they
>> couldn't agree on a common configuration!
>>
>> There are also some differences in the way vendors have chosen to
>> implement the Standard which can lead to... "frustration" (to put
>> it in practical terms  :> ).
>>
>> In the past, I've resorted to manually configuring the interfaces
>> on the various hosts -- just to take one more set of variables out
>> of the equation (it also *seems* like this brings the interface up,
>> quicker?).
>>
>> [You/I shouldn't *have* to do this but I have more fun things to
>> work on than screwing around with what should effectively be a
>> virtual *wire*!]
>>
>> You can also look at your cables to make sure they aren't a part
>> of the problem.  A lot of this stuff is BFM and you'll find that
>> "a wire is not always a wire".  <frown>  Or, that a 6 ft wire
>> is different than a 25 ft wire!
>>
>> To help with your troubleshooting, get out a pad of paper and
>> make notes of what you are doing, what order, and the results
>> you obtain.  Helps improve reproducibility if you know *you*
>> aren't a variable.
>>
>> E.g., you might notice that plugging a cable *after* the switch
>> is powered gives you different results than doing so *before*
>> (as well as whether or not the host on the other end happens to
>> have brought up its interface, already).
>>
>> Cheap things to try are bringing interfaces down and back up,
>> manually configuring them, replacing cables, etc.
>>
>> Good luck!  As I say, I wimped out and just forced things to
>> run the way I wanted them to run...  :-/
>>
>> ______________________________**_________________
>> Tucson Free Unix Group - tfug at tfug.org
>> Subscription Options:
>> http://www.tfug.org/mailman/**listinfo/tfug_tfug.org<http://www.tfug.org/mailman/listinfo/tfug_tfug.org>
>>
>>
>
> ______________________________**_________________
> Tucson Free Unix Group - tfug at tfug.org
> Subscription Options:
> http://www.tfug.org/mailman/**listinfo/tfug_tfug.org<http://www.tfug.org/mailman/listinfo/tfug_tfug.org>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tfug.org/pipermail/tfug_tfug.org/attachments/20130313/acfe21a7/attachment-0002.html>