SUN Ultra 40 M2 NIC Problem

[rapulley]

I am not using the nvidia integrated NIC's, they simply do not work with the forcedeth module. I installed a Sun NIC (sunhme driver) into the box and was working great on F6. I reinstalled today with F8 and the sunhme nic seemed to be working. I was able to connect to network services during install time. However when rebooting, the OS reported that eth0 (sunhme) did not seem to be present ("sunhme device eth0 does not seem to be present, delaying initialization.") At this point kudzu brings up the sunhme card up as eth1 and obtains a DHCP address, network is working. Subsequent reboots the problem starts over ("sunhme device eth1 does not seem to be present, delaying initialization.") and eth2 is brought up, rebooting eth3 comes up and so on. If I disable kudzu, sunhme fails alltogether. I am using the latest distro from fedora, no xen. Any ideas would be appreciated.

[rapulley]

I found the problem but am really not sure why the OS is behaving this way. The issue was that the initial install put the correct MAC in the udev rule (70-persistent-net.rules). When the system is rebooted the mac is randomly generated and no device is found. This would work if the udev rule was also updated but that is not happening. So the system now uses kudzu to add a new device which generates a different mac (not using the hardware address imbedded on the hardware device) that do not match. The issues occured on each reboot.

The fix was simple after identifying the problem. Matching the MAC in the ifcfg-eth0 and corresponding line un the udev rule fixes thes problem I was experiencing. Since I am not seeing anything out there on this i am guessing it is SUN Ultra 40 M2 specific. Any thoughts?

[davesparkes]

Hi,

We've got exactly the same problem as you describe with a Sun QFE generating new MAC addresses on each boot. Running Fedora 9 on a HP Proliant G4 server.

Could you give a bit more detail on how you fixed the problem? I've tried setting the persistent rules file to have the same MAC address as seen generated, but still getting the problem on each boot.

Thanks

[rapulley]

Originally Posted by davesparkes

Hi,

We've got exactly the same problem as you describe with a Sun QFE generating new MAC addresses on each boot. Running Fedora 9 on a HP Proliant G4 server.

Could you give a bit more detail on how you fixed the problem? I've tried setting the persistent rules file to have the same MAC address as seen generated, but still getting the problem on each boot.

Thanks

After matching up the mac addresses correctly I had to disable kudzu. Kudzu was seeing the devices as new hardware and regenerating. I am not sure if kudzu is still part of fedora 9 but it is used in fedora 8.

Let me know how that works out for you.

[davesparkes]

Thanks for the reply, it certainly makes sense. Unofrtunately, it doesn't look as though kudzu is part of Fedora 9, so I'm having a look around the HAL documentation to see if I can disable the fresh hardware scanning on boot (it's not critical for our purposes)

[davesparkes]

Thanks for the reply, it certainly makes sense. Unfortunately, it doesn't look as though kudzu is part of Fedora 9, so I'm having a look around the HAL documentation to see if I can disable the fresh hardware scanning on boot (it's not critical for our purposes)