Brian Long wrote:
On 6/1/07, Shabazian, Chip <Chip.Shabazian@xxxxxxxxxxxxxxxxx> wrote:
This is a problem for us as well. The most reliable solution I've been
given (but haven't had time to try) is to bust open your initrd and
change
the entry for your HBA to "ignore" in the /modules/pcitable file.
There are a number of options that are SUPPOSED to work, like latefcload,
but it hasn't worked for us.
latefcload was added in a fairly recent update for RHEL 3, but I don't
know if it's in RHEL 4. "nostorage" is another option only for RHEL 4
(and probably RHEL 5, but I have not tested it). On my previous team,
we specified "nostorage" for all servers and then ks.cfg contained
"device scsi cciss" for HP. This made sure we didn't overwrite any
SAN storage.
latefcload seems to be there at least since RHEL4u3, and it's
there in RHEL5. we couldn't make much use of it, though, except
in conjunction with ignoredisk in RHEL5. the stubs seem to be
there for ignoredisk in the anaconda that comes with RHEL4, but
it is silently ignored.
using noprobe/nostorage hasn't been an option for us either,
since we have a single kickstart for all of our machines instead
of one per hardware type. haven't taken the time to see about
writing a %pre script which would look at the available storage
devices from a known list and generate the proper device line
for the ks.cfg. besides, it makes sense to let anaconda probe
the hardware since it's pretty good at doing so accurately.
what i'd *really* like to see is the ability to use the device
command to ignore certain devices instead of specify, i.e. "device
scsi qla2xxx ignore" which would load all of the probed scsi drivers
except for qla2xxx. any anaconda developers out there listening...?
chris
