Joe Polcari wrote:
This should help for your last point. I have a %pre script that runs in my kickstarts that fixes the enumeration problem of disks if you need it.
This is appreciated. It's something to add to my bag of tricks. Still, this punctuates the point. Why do we need so many workarounds? Because the kernel's behavior is stupid.
The same could be done for the Ethernet. I have the same problem when you don't want to use the onboard ports and you need the PCI ports to be first.
What I want -- aside from non-stupid behavior -- is a workalike for the Solaris ifconfig plumb and unplumb commands, something that sticks in the kernel configuration for the node. "ifconfig ethX unplumb" and it's gone, invisible, unavailable, forever not there until I "ifconfig ethX plumb" to reactivate it.
For the disks - especially when you have hardware raid and you want the disks in a specific order - you have to load the driver modules in the order you want the disks seen.
This only works if you have different RAID controllers. If you have identical controllers or many disks on a single JBOD interface then there is no way to control the enumeration order. You have to fall back on hacks like volume labels and UUIDs.
-- Rich P. _______________________________________________ Discuss mailing list [email protected] http://lists.blu.org/mailman/listinfo/discuss
