On Mon, 2010-06-21 at 13:05 +0100, John Haxby wrote: > On 21 June 2010 11:44, Masopust, Christian > <[email protected]> wrote:
> I have a server with 2 onboard network-devices (tg3) and one > additional network card (e100). > Everytime there is a kernel update the order of these devices > gets corrupt. So I tried to add > the following "rules-file" to /etc/udev/rules.d: > The RHEL5 way to do this is to put the HWADDR for the devices > in /etc/sysconfig/network-scripts/ifcfg-ethN — the start up scripts > will sort out the device names. Provided there is an HWADDR in there, which there should normally be. Some sites remove it, in order to deploy standard images where the details will vary with the systems used at any given time. To mitigate against these problems, in RHEL5 we did several things under the hood to attempt to ensure consistent device enumeration, so if the kernel is revved, it should normally not affect the order devices are detected. In RHEL6, I wrote a specific udev ordering patch that attempts to ensure we have called modprobe in a specific order, again for the same goal. In the longer term, there is an industry standard being worked on for device to PCI slot mapping that we can use instead of the hardware address on those systems where the address is not desired. Jon. _______________________________________________ rhelv5-list mailing list [email protected] https://www.redhat.com/mailman/listinfo/rhelv5-list
