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

pesky network connectivity on Dell PowerEdge 1850


I am running Debian 3.1 (Linux kernel 2.4.27-2-686-smp) on a Dell PowerEdge 1850 with two Xeon processors and two onboard Intel PRO/1000 network interfaces. I am experiencing a very pesky problem wherein I periodically lose connectivity to network services on that host.

Currently I am just serving SSH and HTTP, and sometimes, after half an hour or so, I am suddenly unable to GET a document over HTTP or connect by SSH. Sometimes services are unavailable for only a minute, and sometimes five minutes, and sometimes several hours. There is nothing in the logs indicating that the NIC went down and back up during this time, and sometimes the interface is pingable (though not always). Sometimes I can connect from one host but not another (but not always). It doesn't appear to be trying to renegotiate the speed and duplex--at least there are no up/down messages in /var/log/messages.

I have tried several different network cables, both onboard NIC's, and two different switches. I recompiled the driver from Intel's source (http://downloadfinder.intel.com/scripts-df-external/confirm.aspx?ftpDown=ftp://aiedownload.intel.com/df-support/9180/eng/e1000-6.1.16.tar.gz&agr=N&ProductID=983&DwnldId=9180&strOSs=All&OSFullName=All%20Operating%20Systems&lang=eng) I am testing from two computers: a Solaris computer attached to the same switch and a remote FreeBSD machine connecting via the internet.

I am running this kernel:

Linux rescomp3 2.4.27-2-686-smp #1 SMP Mon May 16 16:55:31 JST 2005 i686 GNU/Linux

I installed it from this package:


Which I downloaded from via apt-get from:

deb http://mirrors.kernel.org/debian/ stable main

Syn cookies are enabled in the kernel but not in use:

% cat /proc/sys/net/ipv4/tcp_syncookies

Any ideas or suggestions would be much appreciated.


"I'm not exactly a hostage to traditional journalistic ethics, but I didn't understand how something made up could be 'just as true' [as something that is true]".

Phil Greenspun, "The Book Behind the Book"

Reply to: