[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Kickstart via ks=hd not working



On Wed, Dec 17, 2008 at 03:30:55PM -0500, Kyle Powell wrote:
> Phil Oester wrote:
> > When using ks=hd: on the command line, I get an error that the ks.cfg
> > file can't be found, but if I hit enter on the same path, it works the
> > second time.  It seems to be due to the scsi drives not yet being ready
> > when kickstart first looks for the file.
> 
> Looks like a bug, but not necessarily in anaconda. Shouldn't the drive be ready
> before POST completes? Perhaps you have a BIOS option enabled that tells POST to
> complete even though the drive isn't ready (fast POST or something like that)?
> I'm unable to duplicate this on my hardware. I don't know of any parameters that
> would tell anaconda to wait. The only thing I can think of is passing multiple
> (MANY!) duplicate ks parameters. Anaconda will try them all (well, I've only
> tried it with < 4 so there may be a limit, and I've never tried passing multiple
> ks parameters with the same value either). So you could try:
> 
> ks=hd:sda1:/ks/ks.cfg ks=hd:sda1:/ks/ks.cfg ... ks=hd:sda1:/ks/ks.cfg
> 
> and see if the drive becomes ready before anaconda reaches the end of the list.
> I know that's far from elegant, but it's the best I can come up with.

That didn't work, but I found a solution.  The problem is that Fedora kernels
now set SCSI_SCAN_ASYNC=y, so there isn't enough time in the boot process for these
RAID disks to come up.  I simply added "scsi_mod.scan=sync" to the kernel command
line to override the ASYNC, and it worked fine.

Phil


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]