Strange network behaviour

[Vincent Zee]

>>Vincent Zee <basics at zenzee.cistron.nl> writes: 
>
>> On 09 Apr 2003 16:39:03 -0400, Lowell Gilbert wrote: 
>> > Vincent Zee <basics at zenzee.cistron.nl> writes: 
>> > 
>> >>  I am really at a loss at the moment. 
>> >>  
>> >>  I have a 100baseT network at home and all connected machines have 
100Mb 
>> >>  network cards. 
>> >>  When I upload mp3 files from my ibook to the FreeBSD server the speed 
>> >>  drops to 100kb/s. 
>> >>  
>> >>  This only happens between these two particular machines. These same 
>> >>  machines connected to other machines give the expected transmission 
>> >>  speeds. I checked cables, switches and router but found nothing 
>> >>  unexpected. 
>> >>  
>> >>  Any hints, ideas or even solutions are most welcome. 
>> >>  
>> >>  The FreeBSD machine is running 4.7 and the ibook 10.2.4 
>> > 
>> > This sounds a lot like a duplex mismatch. 
>> > Check the collision count. 
>> 
>> Hi Lowell, 
>> 
>> thank you for your answer(:-)) 
>> 
>> Here is the ifconfig output for the nic in the freebsd machine: 
>> 
>> dc0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500 
>>         inet 192.168.1.10 netmask 0xffffff00 broadcast 192.168.1.255 
>>         inet6 fe80::248:54ff:fe12:c767%dc0 prefixlen 64 scopeid 0x1 
>>         ether 00:48:54:12:c7:67 
>>         media: Ethernet autoselect (100baseTX <full-duplex>) 
>>         status: active 
>> 
>> Here is the ifconfig output for the nic in the ibook: 
>> 
>> en0: flags=8863<UP,BROADCAST,SMART,RUNNING,SIMPLEX,MULTICAST> mtu 1500 
>>         inet6 fe80::20a:95ff:fe67:2460%en0 prefixlen 64 scopeid 0x4 
>>         inet 192.168.1.3 netmask 0xffffff00 broadcast 192.168.1.255 
>>         ether 00:0a:95:67:24:60 
>>         media: autoselect (100baseTX <full-duplex>) status: active 
>>         supported media: none autoselect 10baseT/UTP <half-duplex> 
>> 10baseT/UTP <half-duplex,hw-loopback> 10baseT/UTP <full-duplex> 
>> 10baseT/UTP <full-duplex,hw-loopback> 100baseTX <half-duplex> 100baseTX 
>> <half-duplex,hw-loopback> 100baseTX <full-duplex> 100baseTX <full- 
>> duplex,hw-loopback> 
>> 
>> 
>> Here is the output of netstat -i during a transmission of a 14 MB file. 
>> 
>> -bash-2.05b$ netstat -i 
>> Name  Mtu   Network       Address            Ipkts Ierrs    Opkts Oerrs 
>>  Coll 
>> dc0   1500  <Link#1>    00:48:54:12:c7:67  5734078     0  5911495     0 
>>     0 
>> dc0   1500  192.168.1     freeserv         5728716     -  5910566     - 
>>     - 
>> dc0   1500  fe80:1::248 fe80:1::248:54ff:        0     -        0     - 
>>     - 
>> lp0*  1500  <Link#2>                             0     0        0     0 
>>     0 
>> faith 1500  <Link#3>                             0     0        0     0 
>>     0 
>> lo0   16384 <Link#4>                           375     0      375     0 
>>     0 
>> lo0   16384 ::1         ::1                      0     -        0     - 
>>     - 
>> lo0   16384 fe80:4::1   fe80:4::1                0     -        0     - 
>>     - 
>> lo0   16384 your-net      localhost            375     -      375     - 
>>     - 
>> ppp0* 1500  <Link#5>                             0     0        0     0 
>>     0 
>> sl0*  552   <Link#6>                             0     0        0     0 
>>     0 
>> 
>> /\ 
>> Vincent 
>

>Okay, so there are no collisions on the FreeBSD side. 
>Check whatever it's plugged into (a switch?). 
>If *that* thinks the link is half duplex while FreeBSD thinks it's 
>full duplex, you've found your problem.

--------------------------------------------------

Mike wrote:

>Hi: 
>
>I'm experiencing the same issue from a PowerMac 733 with OS 10.2.4 = 
>(which 
>is, to my mind, pretty FreeBSD like) to several different FreeBSD = 
>machines 
>running 4.5 and 4.7 respectively.  I know there aren't any duplex = 
>problems 
>and I'm not experiencing the issue when I go from either the Mac or = 
>FreeBSD 
>boxes to other hosts. 
>
>In doing a TCP dump it looks like packet loss, although it only occurs 
>between these hosts.  The ack is sent by the Mac but the BSD box doesn't = 
>see 
>it, so the Mac re-acks (up to 4 times usually) before the BSD box acks = 
>the 
>ack at the end of its Window.  Everything slows to a crawl until that = 
>one 
>dropped packet is acked. 
>
>Hrmm.   Has anyone else seen this?  I'm thinking it's an issue on the = 
>Mac 
>side, but who knows. 
>
>Mike
>


Hi Lowell and Mike,

sorry for the late reply but the ibook in question stopped working a few days 
ago and I lost all my mail as a result.
The switch saw the connection as 100Mb and full duplex so it wasn't the 
switch.
Mike is describing my situation exactly. So now I really think that it is a 
Mac specific thing.
Thanks Mike for letting me know I'm not crazy (;-))

Vincent