Network stability problems on R710 servers and BCM5709
n.w.bridges at isc.keele.ac.uk
Sun Sep 27 16:31:17 CDT 2009
On Sun, 2009-09-27 at 17:18 +0200, ?ukasz Jagie??o wrote:
> 2009/9/27 Evangelos Souglakos <evansg at otenet.gr>:
> > Hello,
> > We are experiencing network problems on i/o and network loaded R710
> > Poweredge servers. Network connectivity dies after some time.
> > The systems needs to be powered down to bring the NICs back to life.
> > BIOS Information
> > Vendor: Dell Inc.
> > Version: 1.1.4
> Same problem at 1.2.6 ?
> ?ukasz Jagie??o
I've seen similar problems with R610, but in particular when bonding eth0
and eth1 at 1 Gb connected to a Cisco. I've seen situations where I had a
live ssh connection to a server which was still active but a new ssh
connection would hang after prompting for and accepting a user name and
password. The response from the 'live' session would be very slow. Flood
pings would see a high packet loss. Sometimes executing 'ifdown eth1'
revived the session temporarily. The weirdest situation was where the
server was totally unresponsive to ssh connections but happily allowed me
to connect to the Dell management server on port 1311, login and reboot.
The 1.2.6 BIOS did not help, at least not completely. I downloaded the
latest netxtreme2-5.0.17-1.dkms from Dell support, though I then needed a
newer dkms (still on Dell, but not with the patches listed for the R610).
This bnx2 driver (1.9.20b according to modinfo) is now in the
2.6.18-164.el5 kernel, and I now seem to have stability. I'm going to
stress test it a bit over the next day or two, but would welcome other
reports on the subject.
Norman Bridges n.w.bridges at isc.keele.ac.uk
Systems Manager Keele University
More information about the Linux-PowerEdge