Re: A small problem with e1000 Net driver.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 2004-04-21 at 06:01, GianPiero Puccioni wrote:
> Hi,
> 
> I have some problems with a machine (Fujitsu dual Xeon) with an on
> board Intel 82540 network card.
> This is what kudzu reports:
> 
> device: eth0
> driver: e1000
> desc: "Intel Corp.|82540EM Gigabit Ethernet Controller"
> 
> The problem is that it takes a long time to start and the mount of some
> NFS directories fails at boot. I tried to "ping" the machine when it
> boots and the first ping I get back is when the login screen is already
> on. If I try a network stop/start it takes 18/20 seconds before the
> network works, even if I get an [OK] almost immediately.
> 
> 
> The machine is connected to a Cisco switch and I thought that the problem
> could be negotiating the speed so I tried to force eth0 to 100M FD with
> "options Speed=100 Duplex=2" in modules.conf but it made it even slower,
> and it gave an error about "Speed, AutoNeg and MDI-X" being
> incompatible(??).
> Only thing I can think is to put a delay in rc.local and mount the NFS
> directory there, but I hope there is a better solution.
> 
> Additional info,"dmesg" has this:
> 
> Intel(R) PRO/1000 Network Driver - version 5.1.13-k1
> Copyright (c) 1999-2003 Intel Corporation.
> divert: allocating divert_blk for eth0
> eth0: Intel(R) PRO/1000 Network Connection
> divert: freeing divert_blk for eth0
> ip_tables: (C) 2000-2002 Netfilter core team
> Intel(R) PRO/1000 Network Driver - version 5.1.13-k1
> Copyright (c) 1999-2003 Intel Corporation.
> divert: allocating divert_blk for eth0
> eth0: Intel(R) PRO/1000 Network Connection
> ip_tables: (C) 2000-2002 Netfilter core team
> e1000: eth0 NIC Link is Up 100 Mbps Full Duplex
> 
> I don't know anything about this "divert_blk" is this a problem? 
> And this is only the first time, with a "network restart" only the last
> two lines appear, but the delay is still present.
> 
> Any help?

We see this problem too.  I do not believe it is Fedora related.  Like
you we also use Cisco switches.  We have seen numerous cases where Cisco
Catalyst switches take on the order of 20-25 seconds to autonegotiate a
connection (Windows, Linux, and Cisco to Cisco).  We have not, however
found a solution.  You are on the right track with turning off
autonegotiation, but you probably should do it on the PC and the switch.

Instead of the speed and duplex options you might try the AutoNeg=0x08. 

I do not have switch at hand, but IIRC, there are Cisco switch
configuration options which may help too (maybe spanning tree, but it's
been a long time).

Bob...



[Index of Archives]     [Current Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Yosemite News]     [Yosemite Photos]     [KDE Users]     [Fedora Tools]     [Fedora Docs]

  Powered by Linux