PowerEdge 2650 & Redhat 7.3

Brian K. Jones jonesy at CS.Princeton.EDU
Wed Aug 14 07:42:00 CDT 2002


Has anyone heard anything from Dell about these issues?  Is there a
solution?  Is this a known problem without a solution?  Knowing this
would save me some research and 'google-time'.



On Wed, 2002-08-14 at 08:32, Seth Mos wrote:
> At 07:52 14-8-2002 -0400, Brian K. Jones wrote:
> >Actually, I think as far as hardware goes, 7.3 wins here.  However, I'm
> >having some as yet unanswered issues with 7.3 on the 2650 - mainly
> >revolving around the snmp daemon throwing wrong values for the Broadcom
> >5701 10/100/1000 cards, and some NFS issues that were fixed with a
> >kernel upgrade (via RHN).  I'm still debugging and testing, so I'll let
> >you know if other stuff comes up.
> 
> I noticed the SNMP problem but I don't know if the broadcom developers know 
> about them. Both the tg3 and the bcm5700 driver incorrectly report the 
> amount of octets. Something in the order of 1:100.
> Instead of reporting 20Mbit it would report something like 300Kbit in mrtg.
> 
> After replacing the cards with Intel e1000 cards all our traffic stats went 
> back to normal and we were able to judge if a machine had been busy or not.
> 
> >Just so you know, in spite of net-snmp's inability to talk to the
> >Broadcom NICs, several iperf tests are proving them to be quite decent!
> 
> The cards were a tad faster then the e1000 on the same hardware here 
> (400Mbps for the intel vs. 500Mbps for the broadcom). However, a crashing 
> bcm5700 driver and a card that stops working from one second to the other 
> is no good in a server.
> 
> Cheers
> 
> --
> Seth
> It might just be your lucky day, if you only knew.
-- 

Brian K. Jones
System Administrator
Dept. of Computer Science, Princeton University
http://www.linuxlaboratory.org
jonesy at cs.princeton.edu
Voice: (609) 258-6080




More information about the Linux-PowerEdge mailing list