&boot-installer-intro-cd.xml; To boot the installer from hard disk, you must first download and place the needed files as described in . If you intend to use the hard drive only for booting and then download everything over the network, you should download the netboot/debian-installer/i386/initrd.gz file and its corresponding kernel. This will allow you to repartition the hard disk from which you boot the installer, although you should do so with care. Alternatively, if you intend to keep an existing partition on the hard drive unchanged during the install, you can download the hd-media/initrd.gz file and its kernel, as well as copy a CD iso to the drive (make sure the file is named ending in .iso). The installer can then boot from the drive and install from the CD image, without needing the network. For LILO, you will need to configure two essential things in /etc/lilo.conf: to load the initrd.gz installer at boot time; have the vmlinuz kernel use a RAM disk as its root partition. Here is a /etc/lilo.conf example: image=/boot/newinstall/vmlinuz label=newinstall initrd=/boot/newinstall/initrd.gz root=/dev/ram0 append="devfs=mount,dall ramdisk_size=12000" For more details, refer to the initrd 4 and lilo.conf 5 man pages. Now run lilo and reboot. The procedure for GRUB is quite similar. Locate your menu.lst in the /boot/grub/ directory (sometimes in the /boot/boot/grub/), add the following lines: title New Install kernel (hd0,0)/boot/newinstall/vmlinuz root=/dev/ram0 ramdisk_size=12000 initrd (hd0,0)/boot/newinstall/initrd.gz and reboot. If the boot fails, you can try adding devfs=mount,dall to the kernel line. Note that the value of the ramdisk_size may need to be adjusted for the size of the initrd image. From here on, there should be no difference between GRUB or LILO. Let's assume you have prepared everything from and . Now just plug your USB stick into some free USB connector and reboot the computer. The system should boot up, and you should be presented with the boot: prompt. Here you can enter optional boot arguments, or just hit &enterkey;. In case your computer doesn't support booting from USB memory devices, you can still use a single floppy to do the initial boot and then switch to USB. Boot your system as described in ; the kernel on the boot floppy should detect your USB stick automatically. When it asks for the root floppy, simply press &enterkey;. You should see &d-i; starting. You will have already downloaded the floppy images you needed and created floppies from the images in . To boot from the installer boot floppy, place it in the primary floppy drive, shut down the system as you normally would, then turn it back on. For installing from an LS-120 drive (ATAPI version) with a set of floppies, you need to specify the virtual location for the floppy device. This is done with the root= boot argument, giving the device that the ide-floppy driver maps the device to. For example, if your LS-120 drive is connected as the first IDE device (master) on the second cable, you enter linux root=/dev/hdc at the boot prompt. Installation from LS-120 is only supported by 2.4 and later kernels. Note that on some machines, Control Alt Delete does not properly reset the machine, so a hard reboot is recommended. If you are installing from an existing operating system (e.g., from a DOS box) you don't have a choice. Otherwise, please do a hard reboot when booting. The floppy disk will be accessed, and you should then see a screen that introduces the boot floppy and ends with the boot: prompt. Once you press &enterkey;, you should see the message Loading..., followed by Uncompressing Linux..., and then a screenfull or so of information about the hardware in your system. More information on this phase of the boot process can be found below in . After booting from the boot floppy, the root floppy is requested. Insert the root floppy and press &enterkey;, and the contents are loaded into memory. The installer program debian-installer is automatically launched. &boot-installer-intro-net.xml; There are various ways to do a TFTP boot on i386. It could be that your Network Interface Card or Motherboard provides PXE boot functionality. This is a Intel re-implementation of TFTP boot. If so you may be able to configure your BIOS to boot from the network. It could be that your Network Interface Card provides TFTP boot functionality. Let us (&email-debian-boot-list;) know how did you manage it. Please refer to this document. The etherboot project provides bootdiskettes and even bootroms that do a TFTPboot. When the installer boots, you should be presented with a friendly graphical screen showing the Debian logo and the boot prompt: Press F1 for help, or ENTER to boot: At the boot prompt you can either just press &enterkey; to boot the installer with default options or enter a specific boot method and, optionally, boot parameters. Information on available boot methods and on boot parameters which might be useful can be found by pressing F2 through F7F8. If you add any parameters to the boot command line, be sure to type the boot method (the default is linux) and a space before the first parameter (e.g., linux debconf/priority=medium). If you are installing the system via a remote management device that provides a text interface to the VGA console, you may not be able to see the initial graphical splash screen upon booting the installer; you may even not see the boot prompt. Examples of these devices include the text console of Compaq's integrated Lights Out (iLO) and HP's Integrated Remote Assistant (IRA). You can blindly press F1 In some cases these devices will require special escape sequences to enact this keypress, for example the IRA uses Ctrl F , 1. to bypass this screen and view the help text. Once you are past the splash screen and at the help text your keystrokes will be echoed at the prompt as expected. To prevent the installer from using the framebuffer for the rest of the installation, you will also want to add debian-installer/framebuffer=false to the boot prompt, as described in the help text.