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

Re: Attempting customized network install, including ks.cfg



Jeremy Katz (katzj redhat com) said:

> On Thu, 2002-06-06 at 11:05, rpjday wrote:
> > On Thu, 6 Jun 2002, Chad M. Stewart wrote:
> > > (The 7.3 installer is broken if you mount the ISOs as say disc1/ disc2/
> > > disc3/ - should be fixed in next version.)
> > 
> > and that was my next question -- so it's true that net-based ISO installs
> > are broken?  bummer.  i just checked bugzilla and saw nothing about this.
> > will anyone submit a bugzilla report?
> 
> The http/ftp kickstart case unfortunately is... if I find some spare
> time, I might backport the patch and throw up an updated bootnet.img
> with the fix in it.  nfsiso works fine, though, and the ftp/http iso
> loopback mounted stuff works interactively.  (Bug was already filed)

On a slightly related note; on RH7.3 when I attempt to kickstart via PXE
or NFS kickstart w/ bootable media I get failures...

1) PXE kickstart:

the system boots and gets an IP from dhcp; pulls down the vmlinuz and
initrd.img... the pxelinux.0 and hex file for the dhcp-assigned IP located
in pxelinux.cfg directory specifies the location where the ks.cfg for the
node being build is found.

Therefore, the files in pxelinux.cfg which have contents like:

DEFAULT pxelinux.cfg/vmlinuz
APPEND ks=nfs:192.168.0.230:/tmp/build_daemons/LC1/ks.cfg initrd=pxelinux.cfg/initrd.img
PROMPT 1
TIMEOUT 30

that are supposed to pass the APPEND section to the installer are
seemingly NOT passing what they should.. or th einstaller is puking on
it.  Either way has anyone else seen this?


2) NFS kickstart with boot media:

The system is booted with a bot floopy or boot cdrom that contains the
bootnet.img; the installer then gets an IP via dhcp; mounts /kickstart and
pulls down the <assigned-ip>-kickstart and then runs anaconda... all this
works!  

The problem comes with scsi based systems where after anaconda is run with
the ks.cfg that it just pulled off of nfs; a dialog appears that says it
can't find a disk to install onto (namely sda) and /tmp/ does NOT contain
sda.

Has anyone seen this?

Both these methods worked like a champ with anaconda-7.2; I'm getting very
non-deterministic results with anaconda-7.3.

The only ks method that works reliably on RH7.3 for us is to use a cdrom
kickstart with something like ks=cdrom:/path/to/ks.cfg in syslinux.cfg

All this is pretty frustrating, thanks for any insight you might have!

Mike






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