Re: Installation failed - and failed again...
On 2013/3/3 6:25 AM, Brian wrote:
On Sat 02 Mar 2013 at 19:56:43 -0500, Mark Filipak wrote:
On 2013/3/2 3:36 PM, Brian wrote:
Your success was not a result of running the text-based Debian
Installer. Forget about it. Try a bit of lateral, or even vertical,
thinking. You may not realise it but substituting the USB stick you
initially used for a USB hard disc is the crux of the matter.
Observation: I'm struck by how some (many?) on this list tell me what
I did (like I was asleep) and they're wrong. Please, pay attention...
I did not make a substitution.
Ok. You are using USB flash/USB flash drive/USB hard disc/USB drive
interchangeably to refer to the same device? Its good to have that
clarified.
No, Brian. I'm *not* using terms interchangeably. I'm *not* being sloppy. When I refer to a USB flash drive I mean a USB flash drive (also called a thumb drive by some people and plugged into one of my laptop's USB 2.0 ports), and when I refer to a USB hard drive mean a USB hard drive (a hard drive in a USB carrier and plugged into another of my laptop's USB 2.0 port via a cable).
I've thrown the thread away, but if you review my previous posts I'm pretty confident you will find that I've been absolutely consistent.
Source: USB flash drive booting Debian Live.
Target: USB hard drive.
Prior to attempting to install to the USB hard drive, I tried to install to an 8-GB USB flash drive - that's right: initially there were 2 USB flash drives. I differentiated between them this way: source: 1-GB USB flash drive booting Debian Live, target: 8-GB USB flash drive. The reason I switched the target from an 8-GB USB flash drive to a USB hard drive is that someone said I couldn't install to a USB flash drive because of the flash structures and that I should switch to a hard drive because the MBR will be easier to deal with. I thought that suggestion was bogus, but I went along with it because I had a spare USB hard drive and because I wanted to move past that roadblock, even if bogus.
1 - I installed Debian Live on a USB flash drive.
2 - I booted to the USB flash drive.
This is the 1GB drive. Correct?
You got it! Yes.
3 - I invoked the GUI version of the Debian Installer and attempted to
install Debian to a USB hard disk. - Both GRUB and LILO installation
failed (GRUB is automatically tried, and fails, LILO is tried
manually).
The install medium was the 8GB drive. Correct?
By this time the target was the USB hard drive... mmmm... I think it's about 50 GB but that's not really important.
4 - I invoked the text-based version of the Debian Installer targeted
to the same USB hard disk and it succeeded.
Of course, between steps 3 & 4 I had to reboot back to Windows and
delete the USB hard disk partitions because deleting them inside the
GUI Debian Installer didn't work, so I guess it actually has more
bugs, eh?
No, no more bugs;
But they're so tasty! Can't I eat just one more?
you have had your fair share up to now. Don't be
greedy; leave some for other people. You've used the very fuzzy term
"didn't work", but whatever you did triggered the same bug you had
already met.
Yes, same bug: GRUB failed to install automatically. Then, when I attempted to manually install LILO, it failed also. Note: when I write that I attempted to manually install LILO, I don't mean that I opened a terminal window and used the command line - I don't know how to do that - I mean I double-clicked LILO from the main menu.
The point is: in both cases (unsuccessful & successful) the
source of the install was Debian Live on a USB flash drive and the
target of the install was a USB hard disk.
No, the two cases are definitely not identical.
Case1 (Unsuccessful)
--------------------
The 8GB drive had previously had Debian Live installed to it. (This is
clearly described in your first post). The target drive would contain
information about an iso9660 filesystem. GRUB is designed not to
install to a drive when it detects an iso9660 filesystem is present on
it. [1]
What you have written above is not correct. I *had* installed Debian Live+Gnome on the 8-GB USB flash drive *but* I didn't use it. I installed Debian Live+LXDE on the 1-GB USB flash drive and *used* *that* as the source. The target *was* going to be the 8-GB USB flash drive (overwriting the existing Debian Live installed on it), but, when that didn't work, I switched the target to the USB hard drive as suggested by someone.
To be comprehensive, while booted to Debian Live+LXDE (1-GB USB flash drive), I attempted (several times) to delete the partitions on the 8-GB USB flash drive (using the installers menu, selecting 'delete partition'). I know the installer was trying to delete the partitions because it would fail - I'd have to quit the installer and open up a file browser, unmount the existing volumes, then rerun the installer and attempt to delete the partitions again. When that didn't work, I'd quit, boot to Windows, and delete the existing partitions in Windows. Then I'd boot back into Debian Live+LXDE and try the installation once more. When that still didn't work, I put the 8-GB USB flash drive aside and switched the target to the USB hard drive. From then on, the target was *always* the USB hard drive.
There were many opportunities for failure. Perhaps the partitions were not actually being deleted. Perhaps the volumes were not actually being unmounted. I'd be willing to try to replicate the problem if the folks here will give up their trolling attacks on me and give me some actual help using actual steps that can be carried out: 'do this', then 'do this', then 'do this', etc. (step by step).
Case1 (Successful)
------------------
The 8GB drive had been processed by a Windows partitioning program.
This wipes the information about the iso9660 file system. GRUB will
now install.
I reiterate:
1 - I installed Debian Live on a USB flash drive.
2 - I booted to the USB flash drive.
3 - I invoked the GUI version of the Debian Installer and attempted to install Debian to a USB hard disk. - Both GRUB and LILO installation failed (GRUB is automatically tried, and fails, LILO is tried manually).
4 - I invoked the text-based version of the Debian Installer targeted to the same USB hard disk and it succeeded.
Now, one could reasonably assume that the Debian Installer would unmount any existing partitions on the target drive. Ah but, you might say, the user may want one of the existing partitions. "Yes", I say. "That's true." Well, not being stupid, I was choosing the 'remaining space' (or whatever the phrase is) in the installer or I was deleting partitions (or at least, attempting to delete them). Wouldn't it be much easier for the installer to present the user with the existing partition table and ask the user to mark the partitions to keep and the partitions to delete? Is that too much to ask? Lacking that, wouldn't it make sense that, if the user wants to delete an existing partition, that the installer would unmount it first instead of just failing with a message that the volume is busy? Is that too much to ask?
Is this stuff tested at all? Why does Linux make everything so difficult?
At any rate, I would quit, boot to Windows, delete the existing partitions, boot to Debian Live+LXDE, and try again.
Now that you mention it - what does GRUB report in the syslog when it
fails to install as you described in your very first post? Be daring;
give it a go again, exactly as you related there.
Sigh! How can I read the syslog? Where is it? /var/...blah-blah-blah... or
/usr/...blah-blah-blah... or /sys/...blah-blah-blah... - I'm not a Linux
user. And, since nothing gets written to the USB flash drive (remember: this
Debian Live thinks it's running on CD), how would I find it at all!
http://lists.debian.org/debian-user/2013/03/msg00187.html
You succeeded (which is good) because you changed the install environment.
This neither proves nor disproves anything.
I did not change the install environment. What makes you think I did?
You have told us as much above. The state of the 8GB drive for the text
mode install is different from its initial state.
I was not using the 8-GB USB flash drive.
Incidentally, it is never a good idea to change more than one variable
at a time in an experiment. It can (as it has in your case) lead to
an invalid or erroneous conclusion.
I did not change variables. Source: 1-GB USB flash drive. Target: USB hard drive.
[1] For those whose wish to test this assertion:
a) Plug a USB stick in and identify its device name with
dmesg | tail -n 20
b) Write an isohybrid ISO (an amd64 or i386 netinst would do) to the
stick with
cat ISO > /dev/sdX
c) Install GRUB with
grub-install /dev/sdc
d) Re-partition the stick with fdisk and repeat c).
Reply to: