[PLUG] Slowest Full duplex 100MB in the west

Ronald Chmara ron at Opus1.COM
Tue Mar 15 11:12:18 UTC 2005


On Mar 15, 2005, at 2:12 AM, Michael Rasmussen wrote:
> Ronald Chmara wrote:
>>> Errrr? What do you mean?
>>>> jbrelin at robot:~# ifconfig eth0
>>>> eth0      Link encap:Ethernet  HWaddr 00:A0:CC:E5:A5:34
>>>>          inet addr:10.42.42.41  Bcast:10.42.42.255
>>>> Mask:255.255.255.0
>>>>          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>>>>          RX packets:8794083 errors:42726 dropped:0 overruns:0
>>>> frame:63831
>>> ------------------------------>^^^^^^^^^^^^
>> That's not enough to cause the level of losses you were talking
>> about....
> Sure it is, if the errors occured during the ftp process.

'splain, please. That's only 0.4 percent of packets, so unless each bad 
packet triggered a fairly *lengthy* re-negotiation process, I'm not 
seeing how it would drop a 100Mb/s to 8KB/s. Maybe 100Mb/s 
(theoretical) to 30 Mb/s (actual), but 64 Kb/s? Roughly five hundred 
times slower?

> Thus he needs to follow Wil's suggestion to use  mii-tool and
> force the client to auto, then 100/full, then 100/half and test
> the throughput under each scenario.

I do agree that this is a good test. However, I'm used to duplexing 
madness dropping a 100Mb down to 30 Mb/s speeds, but I've never seen 
duplexing issues drop a connection down to 8KB (64Kb) speed. To drop 
that low, I'd guess crappy cables, or *serious* interference, or 
something similar on the physical layer...

-Bop




More information about the PLUG mailing list