I am trying to install SL 6.2 to a host partition on my machine. I have
tried twice and lost the ability to boot my machine at all, including
the installed CentOS 6.2 host partition. (Note that I point out the
term "host" to differentiate it from the VM used to create the install
medium.)
Let me go through the steps I took. First, I created a SL 6.2 VM using
VirtualBox 4.1 and applied all updates from the repos. This VM runs on
my CentOS 6.2 installation, from which I am writing this message today.
That installation went smoothly, no hiccups, etc. I did minimal
configuration to that VM install. That VM does boot successfully,
repeatedly, without incident.
Next, I created a bootable USB drive containing the SL 6.2 liveCD image.
I ran the liveusb-creator from INSIDE the VM. But note that the
physical USB drive lives outside the VM in the real, physical world of
the host system -- is this a problem? It's all solid state media, and
even the data transfer is handled internally from the guest to the host.
I took all the defaults in generating the USB drive. I watched
carefully as the USB drive was written to; the LED blinked while the
write progressed.
I shut down the VM guest, then rebooted the host (hardware) from the USB
drive. I followed all the instructions, taking once again, all the
defaults except to point the install to a specific partition on the
host. This partition formerly housed an older CentOS installation no
longer used, and before that another distro (I forget which now). I
chose to reformat that partition and use that entire partition for the
SL 6.2 install.
At some point after the copying of files, anaconda (python) failed with
an error. The specific error escapes me right now, but after tracking
it down in google and other places, it seems that changing the domain
name in the post-install steps can cause this to happen. So I attempted
a second install.
This time, anaconda did not cause a failure/traceback, and the install
appeared to complete. When the installation was finished, it asks to
reboot (this followed the VM install exactly), and that is what I did.
However, the hardware could not find the grub install.
I tried a few different tricks. The hardware drives on my system are
RAID1, so I tried forcing a boot off the other drives in the RAID set,
to no avail. I then reached for my copy of the ever-ubiquitous
LinuxRescueCD and booted into the install partition, which also did not
work well since the modules and libraries are at different levels, so my
wireless could not start, and so on downhill from there.
Finally, I rebuilt the original grub install from my CentOS partition by
doing the chroot trick or the like and got my CentOS 6.2 back. Whew!
That was frightening.
I would not like to go through all this all over again. Is there
something inherently presumptuous about my approach? Again, I created a
USB install using the standard tools, albeit from a VM. I admit that
there could be some issue with VM USB access and data update, but I
rather doubt it. I haven't had any other problems with USB under
VirtualBox.
I suppose I could try burning a CD, but what if the same thing happens
again? That is, I'd like to avoid some sort of hardware compatibility
problem with my mainboard, etc, and I end up close to losing my entire
system again.
Anyone hazard a guess on where/why this could be a problem? I posted
this to the forum, and one user suggested trying SL 6.1 and rolling it
forward. But he also indicated that PXE install was the best route,
which, if true, would point to some sort of media transparency issue.
Thanks to anyone who can help. I'd like to try out SL on my home system.
|