Answers again.
> I can talk to Solaris boxes fine with 2.2.x. Everything I hear sounds more
> like misnegotiation of duplex and error problems even now.
Can you give me the specifics on your config on linux and solaris (ndd's
parms etc) so I can compare it? Also patch levels on solaris and OS
version would help me compare.
I am positive of the cards being in full-duplex not only he switch says
they are, (Usually Lies cause the autosend code sucks), but mii-diag was
used to force the cards to full-duplex, If I switch to half-duplex on
purpose, the performance drops to below 500kB/sec
>
> > 5 packets were dropped by Linux and had to be re-transmitted. This
> > contributed to about 30% of the elapsed time. Looking at the traces in
> > detail, the receiver isn't using multiple ACKS to indicate a dropped
>
> [We use multiple acks] Not having acks implies we never saw the packets
> in the first place, which means they didnt get through. That indicates
> that either the switch cabling card or card-driver got involved - hence
> duplex issues and the like.
I wish it were that easy, but to remove the switch from the picture I ave
done the test with even 2 computers connected via a flip cable, with
results no different than on a switch.
>
> > Case 2)
> >
> > The Sun had to slow down and wait for the Linux box to empty the
> > receive window 4451 times. This accounted to about 20% of the elapsed
> > time. It indicates the efficiency of the Linux TCP stack is poor when
> > emptying data. A faster CPU would help.
>
> Actually its very good, but it depends on application space to be able to
> handle large amounts of data fast, and that can be an application issue.
>
> What to me is most interesting is - why did a chunk of packets go walkies.
>
> What rev of the eepro100 driver are you using ?
The one in the RH6.2
/var/log/messages says eepro100.c Rev 1.18, 1999/12/29
>
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
