Ahh, this is all automated in the latest systeminstaller package,
shipped in OSCAR 2.3.
It actually does it a different way, all through PXE booting. The
default file (as now) is to load a new image, but it can also create a
config file for the individual node which fails PXE booting, falling
through to a local disk boot. Of course for this to work, the node has
to try PXE booting before the local disk.
There are 3 steps:
1) Change the last line of /etc/systemimager/systemimager.conf to have
"NET_BOOT_DEFAULT = local" (instead of "net"),
2) Start the netbootmond process, i.e.
/etc/init.d/systemimager-server-netbootmond start
(and fix up to run after each reboot), and
3) Reload each of the clients.
This process works by the netbootmond process watching the file
/var/log/systemimager/rsyncd for the "image_completed" string and then
creating a PXE booting config which disables PXE loading for that node
(actually a symlink to
/etc/systemimager/pxelinux.cfg/syslinux.cfg.noboot).
After this, if you want to image a client, you just run
/usr/sbin/mkclientnetboot with the host name. This removes the special
PXE config file for that node, forcing it to reload the image.
I've now been using it for a few weeks and it seems to be fine.
Frank
On Thu, 2003-10-02 at 09:22, Bernard Li wrote:
> Hey Jeremy:
>
> That's a good idea - just in time for me to set up these bad boys :-)
>
> Thanks,
>
> Bernard
>
> Jeremy Enos wrote:
>
> > Bernard-
> > Another suggestion:
> > When I'm rebuilding clusters here, I'll frequently just make the boot
> > order on the clients HD, then PXE. So that way, PXE is only used as a
> > boot device when the HD is not a valid boot device. The trick comes in
> > there... for new HDs, or new machines, the HD is already invalid and I
> > don't have to do anything. For existing machines, it depends on me to
> > mung the boot sector of the HD before rebooting... So- if this fits
> > your needs, you could do it this way. I just run:
> > dd if=/dev/zero of=/dev/hda count=2
> > And that's enough to invalidate the HD as a boot device. (obviously,
> > substitute hda w/ sda for scsi, etc)
> >
> > Jeremy
> >
> > At 02:22 PM 10/1/2003 -0700, Bernard Li wrote:
> >
> >> Hey Sean:
> >>
> >> This is odd - I am actually trying to do what you mentioned in your
> >> message but I noticed that I only have 2 files in my
> >> /tftpboot/pxelinux.cfg:
> >>
> >> [EMAIL PROTECTED] pxelinux.cfg]# ls -l
> >> total 8
> >> -rw-r--r-- 1 root root 189 Sep 29 15:08 default
> >> -rw-r--r-- 1 root root 721 Sep 29 15:08 message.txt
> >>
> >> and this doesn't change throughout an Oscar installation of the nodes
> >> - perhaps I'm doing something wrong here.
> >>
> >> I am actually trying to setup Oscar on these IBM x330 nodes and it's
> >> not as easy setting up the bootup settings as our Blade Center with
> >> all the centralized management utilities. For the x330's I had to go
> >> into BIOS to change the boot sequence individually - not fun ;-)
> >>
> >> Cheers,
> >>
> >> Bernard
> >>
> >> Sean Safron wrote:
> >>
> >>>
> >>> I assume you are trying avoid the situation where you do a network
> >>> install,
> >>> and then when the node reboots, it doesn't reboot from the hard
> >>> drive, but
> >>> does another network install!
> >>> There are various ways to get around this problem, all of which involve
> >>> moving the HEX file out of the way after the install is complete, or
> >>> changing the node's HEX file to boot from the local disk after the
> >>> install
> >>> is complete. One of these techniques is described in the AutoYaST2
> >>> guide
> >>> (http://www.suse.de/~nashif/autoinstall/8.1/html/x789.html#AEN840).
> >>> Even
> >>> if you don't use SuSE, this technique applies to other Linux
> >>> distros. Here
> >>> is the relevant text:
> >>> You can watch the syslog file on your TFTP server and whenever a
> >>> client got its initial
> >>> RAM disk transmitted, you can remove the symlink for that machine
> >>> from the
> >>> pxelinux.cfg directory. This forces the client to load the default
> >>> configuration
> >>> which says: "Boot from local disk!" when it reboots after
> >>> AutoYaST is
> >>> done.
> >>>
> >>> If this is not the situation that you are encountering, then I don't
> >>> understand the problem. It usually takes less than a second for pxe to
> >>> scan through all the combinations of the HEX filename and eventually
> >>> find
> >>> the default problem.
> >>> -Sean
> >>> ---------------------------------------------
> >>> Sean S. Safron - Linux Cluster Development - IBM, Poughkeepsie, NY
> >>> 845-433-8103 (tieline 293-8103) POK/414 1-24
> >>> [EMAIL PROTECTED]
> >>>
> >>>
> >>>
> >>> Kaizaad Bilimorya
> >>> <[EMAIL PROTECTED]> To:
> >>> [EMAIL PROTECTED]
> >>> Sent by: cc:
> >>> [EMAIL PROTECTED]
> >>> [EMAIL PROTECTED] Subject:
> >>> [Sisuite-users] hex ipaddress override
> >>> ceforge.net
> >>>
> >>>
> >>>
> >>>
> >>> 09/30/2003 07:24 PM
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> HiYa,
> >>> Does anybody know of a way to overide the network boot setting so
> >>> when you
> >>> network boot a client, instead of it trying to load,
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT>
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT - last character>
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT- last 2 characters>
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT- last 3 characters>
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT- last 4 characters>
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT- last 5 characters>
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT- last 6 characters>
> >>> pxelimux.cfg/<HEX_IP_OF_CLIENT- last 7 characters>
> >>> pxelimux.cfg/default
> >>> it goes after the
> >>> pxelimux.cfg/default
> >>> right away?
> >>>
> >>> thanks
> >>> kaizaad
> >>>
> >>>
> >>> -------------------------------------------------------
> >>> This sf.net email is sponsored by:ThinkGeek
> >>> Welcome to geek heaven.
> >>> http://thinkgeek.com/sf
> >>> _______________________________________________
> >>> Sisuite-users mailing list
> >>> [EMAIL PROTECTED]
> >>> https://lists.sourceforge.net/lists/listinfo/sisuite-users
> >>>
> >>>
> >>> -------------------------------------------------------
> >>> This sf.net email is sponsored by:ThinkGeek
> >>> Welcome to geek heaven.
> >>> http://thinkgeek.com/sf
> >>> _______________________________________________
> >>> Oscar-users mailing list
> >>> [EMAIL PROTECTED]
> >>> https://lists.sourceforge.net/lists/listinfo/oscar-users
> >>
> >>
> >>
> >>
> >>
> >> -------------------------------------------------------
> >> This sf.net email is sponsored by:ThinkGeek
> >> Welcome to geek heaven.
> >> http://thinkgeek.com/sf
> >> _______________________________________________
> >> Oscar-users mailing list
> >> [EMAIL PROTECTED]
> >> https://lists.sourceforge.net/lists/listinfo/oscar-users
>
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Oscar-users mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/oscar-users
--
ac3
Suite G16, Bay 7, Locomotive Workshop Phone: 02 9209 4600
Australian Technology Park Fax: 02 9209 4611
Eveleigh NSW 1430
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Oscar-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/oscar-users
