> > months ago) throughput between my two boxen on a private ethernet segment
> > has dropped by a factor of over 100 from almost half a MB to just a few KB
>
> half MB is a bit low but ok
Part of that may be due to the fact the the sending end is a 120 Mhz 5x86
machine. The motherboard there is about four or five years old.
>
> > per second. tcpdump shows only a few packets per second and ifconfig
> > shows 1 carrier error about every 2 seconds on the tx side and the same
> > number of frame errors on the rx side. The send queue on the tx side gets
>
> framing/carrier errors are consistent with a hardware problem or low level
> driver problem. The framing is done in hardware, carrier errors are hardware
> and framing on rx errors are hardware level. What devices are you using for
> this ?
One DEC Tulip card in the 5x86 machine and one PCI NE2000 card in 350Mhz
K6 machine.
>
> > filled to capacity and stays there until the transfer is complete. I
> > think this is a network stack problem and not a hardware problem because
> > ping -f gets 0% packet loss (one run lost 10 packets out of 60,000). This
> > only happens with bulk transfer, interactive sessions are perfectly fine.
>
> ping -f is a meaningless test, it generates no long back to back frames,
> and few frames with pessimal ethernet clocking.
>
------------------------------------------------((\))<----------------------
Nate Riffe Duct tape by any other name is just as
inkblot@geocities.com sticky.
-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GCS/IT/M/S/O d-@ s-:- a--->- C++ UL++++BS++>$ P+>+++ L+++>+++++$ W+ N !o
K- w(---)$>-- M-(--) V(--) PS+ PE Y+ PGP>++ t(+)@ 5 X@ R tv>! b+>+++ DI++
D e>++(+++) h r++ y?
------END GEEK CODE BLOCK------
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/