Hi!
I've been chewing on this for three days - now it goes before the experts...
:)
My LAN looks like so:
Machine 1
Dual P2/400/256
somewhere between potato and woody | Win NT 4sp6a
eth0: D-Link DFE-530TX (10/100)
- via-rhine.o from http://www.scyld.com/network/ethercard.html
- to switch
- static ip 192.168.0.1
eth1: 3Com 3C905B (10/100)
- 3c90x.o from 3Com web site
- to cable modem
- DHCP
Machine 2
Celeron/533/128
Win 98SE
eth0: D-Link 530TX (10/100)
- to switch
- static ip 192.168.0.2
Machine 3
P/100/24
potato
eth0: 3Com 3C900 Combo (10)
- 3c59x.o from 2.2.15
- to switch
- static ip 192.168.0.101
The switch is a D-Link DES-1008 (8 port autosensing)
Now: I can ping each machine from itself and all others, so it's basically
working.
Only when I start file transfers it becomes apparent that the transfer rates
are way too low.
Machine1(linux) --> Machine2: max 3MB/sec, typ 500KB/sec
Machine1(linux) <--> Machine3: max 200KB/sec, often only 3KB/sec
Machine3 --> Machine2: once got 600KB
...
Whenever Machine 1 was involved it would output
"Something wicked happened"
with either 0009 or 000a appended to the kernel log
and every console.
There is a paragraph about this message at the site where I got the driver,
but it only lists different error codes.
While NT was a bitch to get working at all, it now gives me (2)-->(1)
6500KB/sec
over "windows networking" (aargh!) which seems to be the maximum (2) is able
to
handle.
Should I blame the via-rhine driver?
Have any of you got any ad-hoc ideas?
What info should I provide?
Are there any good diagnostic programs (besides mii-diag)?
...
Help!
(At least I hope you can help me - I'd have to let go of that
"dump nt, go samba" idea of mine, otherwise.)
Best regards
Christian Pernegger
