Hi, I have a very weird problem which I'm unable to diagnose after trying various things. So this maybe long, please bear with me. I bought a new IBM x3250 server. It has two gigE ports.
lspci :- 0b:00.0 Ethernet Controller : Intel Corp 82574L Gigabit Network Connection 15:00.0 Ethernet Controller : Intel Corp 82574L Gigabit Network Connection lspci -n |grep 0b:00.0 :- 0b:00.0 0200: 8086:10d3 (same for the other one) uname -r :- 2.6.18-194.el5 I installed CentOS 5.5 and configured the interface for dhcp. It fails to connect and just keeps fluctuating. In /var/log/messages/ I see Link is up/down message repeatedly (every second). I can give it a static ip and define a static route, but it does absolutely nothing except self-ping. First I thought it was a driver issue, so I updated to the latest driver for my card (e1000e) first from elrepo.org (didn't work) then from the intel website (didn't work). elrepo btw is not up to date with intel. I called the IBM guys who showed up today with a Windows Server CD today, installed it- same problem there. They called for a new motherboard, put windows 2003 server edition, same problem! They updated the BIOS 3 times.. same problem on all 3. They went back without a solution, collecting the DSA logs (it's an ibm thing i think). They haven't closed the call and are still on it. I then got a new, exact same machine from the vendor, I ran a live cd first (linux mint) and I get a STABLE link, booted windows from the previous installation (hot-swap drives) STABLE link. IBM guys go home, I install CentOS and poof! back to square one! :( Then I tried linux mint, same issue. You can rule out faulty cables or switch ports, I've verified that. Ok, this is really weird, I just found a workaround. By just I mean after I'd typed all of the above email. Just for kicks I put in a dumb 10/100 mbps hub in the path and it works great :) This however, is temporary. Could this be a link negotiation issue? Also, what baffles me is that this problem persisted across 3 motherboards and 2 machines only after CentOS was brought into the picture. CentOS first, followed by other distros. Failed for every one of them. Other OS first (including windows) STABLE link, bring CentOS into the picture, it failed and also for every subsequent OS. I read somewhere that there was a time when the e1000e driver in the kernel was bricking NICs but somebody also told me that code didn't make it into CentOS. Anybody ever come across such a thing? Any help would be much appreciated. Thanks Gurteshwar _______________________________________________ Ilugd mailing list Ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd
