I'll have to look at the iscsi and zfs initramfs hooks, and see if I can model it most concisely on what they currently do. Between the two, I should be able to hack something up.
On Mon, Jan 27, 2014 at 9:46 PM, Stuart Longland <[email protected]> wrote: > On 28/01/14 15:29, Schlacta, Christ wrote: > > iPXE supports loading a kernel and an initrd from https. > > http://lists.ipxe.org/pipermail/ipxe-devel/2012-May/001465.html > > The part I have 0 experience with, and wouldn't know where to start is > > with modifying hooks for initramfs. I am hoping someone has most of this > > hashed out already. > > > > Coincidentally, since ceph can export files over http, and I'm guessing > > over https, it should be fairly simple to simply mount /boot over cephfs > > and also have it exported via https for the boot process. > > Not sure of the full details, but the way I'd approach this would be: > > - Build a ramdisk with busybox, libc, Ceph's `rbd` tool, and any other > needed libraries, kernel modules, etc. > > - Create a shell script /linuxrc with something like the following: > > #!/bin/sh > > # modprobe any drivers you need here; rbd, etc > > # Bring up your network interface; pick up DNS settings, etc > udhcpc eth0 > > # Map the rbd device > rbd map pool/rbd-name > > # Tell mdev to generate the /dev node > mdev -s > > # Now try mounting the device > mount /dev/rbd0 /initrd -o ... # mount options here > > # The following is from the pivot_root manpage: > pivot_root . /initrd > exec chroot . sh -c 'umount /initrd; exec /sbin/init' \ > <dev/console >dev/console 2>&1 > > - Optional: symlink /linuxrc to /init; for initramfs compatibility > - Package up your initrd, then cross your fingers. > > I have no idea if the above will work, but that'd be my approach. > > Regards, > -- > Stuart Longland > Systems Engineer > _ ___ > \ /|_) | T: +61 7 3535 9619 > \/ | \ | 38b Douglas Street F: +61 7 3535 9699 > SYSTEMS Milton QLD 4064 http://www.vrt.com.au > > >
_______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
