Mentioned unsupported SATA RAID software systems.
|Deletions are marked like this.||Additions are marked like this.|
|Line 105:||Line 105:|
|[http://linux.yyz.us/sata/faq-sata-raid.html SATA RAID] Chipset's or PCI/PCIe-card's software RAIDs are not supported yet, or not completely, see dmraid [http://lists.debian.org/debian-boot/2005/11/msg00297.html 1] [http://packages.debian.org/unstable/admin/dmraid 2], if you do not need to dual boot to a MS Windows installed on a such software raid system, please use the debian-installer default raid manager, above mentioned, it is far more tested.
|Line 217:||Line 219:|
(FYI: last I heard the Debian kernel maintainers refused to maintain
a forked tg3, and the upstream kernel maintainers refused to accept any
changes. Currently the non-free version is in Debian, but it will probably have to be removed again, unnecessary though that is --Nathanael Nerode)
Before attempting an installation or reporting a problem, remember to check the [http://www.debian.org/devel/debian-installer/errata errata] page to see if your problem is already well known.
It might also be worthwhile to check [http://www.debian.org/devel/debian-installer/ports-status this page] to see the status of the DebianInstaller on different architectures and with different installation methods.
Finally, if you are experiencing problems with an older version of the installer, please try a more recent version from the [http://www.debian.org/devel/debian-installer/ Debian-Installer home page], as your problem might already be solved.
List of questions
- Hardware is detected at install time but not after reboot.
- Why use DHCP automatically?
- I don't agree that DHCP should run automatically.
- Why not use an existing installer from another distro?
- Will there be a graphical installer?
- Will there be support for software RAID?
- How do I install using LVM?
- Is Serial ATA (SATA) hardware supported?
- How can I add support for a driver for which I have the sources?
- Is the 2.6 version of the Linux kernel supported?
- How to install with boot floppies + netinst.iso mounted as a loop device in the ramdisk from a ext2 partition?
- How can I create boot iso with my own set of packages?
How can I build the DebianInstaller
- After installation, my machine just keeps rebooting!
- Problems with ide-scsi module
- Why do I have to choose twice which mirror to use?
- Why doesn't the installer automatically choose an optimized kernel?
- How about SMP support?
- Why isn't the installed system configured to use the language I chose during installation?
- Why isn't X configured to use the keyboard I chose during installation?
When I run the DebianInstaller, my keyboard doesn't work!
Why isn't my NIC supported by DebianInstaller, I know the "tg3" driver supports it!
- There seem to be errors reading the CD!
How can I install sid (unstable) with DebianInstaller?
- How do I copy d-i (log)files to a remote host?
- How do I use netinst over ppp(oe)?
- What do I do if I can no longer boot Windows after installing Debian?
- Can I use d-i as a rescue system? What pitfalls are there?
- How to use preseeding? Howto use network console with preseeding?
- How to shrink an existing windows installation?
Question 1: The installer detects my ethernet card and loads the proper modules for it, but when I reboot, no drivers are loaded (the card doesn't work).
Answer: This is most probably a known problem with discover-data. The problem will go away once discover-data is upgraded to a more recent version. This should have already happened now, so if you see the problem, let us know.
Question 2: Why does DebianInstaller always use DHCP? I want a static address!
Answer: There are currently two different ways of configuring a static IP setup:
- Run in normal mode and wait when the DHCP step is performed, if it fails you are given the option of configuring a static setup.
- Boot the installer in expert mode ("boot: expert"), and you will be able to configure a static address, and many other things besides. (Press F2 in boot screen for more information.)
Besides you may set your dhcpd to serve a fixed address to your chosen network card (identified by its mac).
Question 3: But DHCP shouldn't run automatically, it breaks on my system/gets the wrong information/is not what I want.
Answer: DebianInstaller is targeting a larger base of users than have historically installed Debian, including users who don't know what DHCP or static IP addresses are. Thus default behavior is chosen to use DHCP.
You can force static network configuration by providing boot parameter netcfg/disable_dhcp=true at the boot prompt. (Press F7 in boot screen for more information.)
Question 4: Why are you doing this? Why not use Anaconda/PGI/whatever? Why reinvent the wheel?
Answer: We're doing this because it's time. Debian has been burdened with a sub-par installer for a full decade, and it's time to change all that. Debian's installer needs to work on more platforms and more types of install media than any other. We need to support installs to S/390 mainframes that have only a teletype console. We need to support installs to m68k boxes that boot from floppies, as well as to netbooting Sun hardware.
And yes, we want to support flashy easy installs on Intel clones booting from CD (and USB sticks). And many more; all the currently targeted installation methods are listed [http://www.debian.org/devel/debian-installer/ports-status here].
PGI can't do this. Anaconda can't do this. Only a system designed from the ground up to be completely modular and highly flexible can do so. DebianInstaller is that installer.
Answer: The DebianInstaller will not be graphical by nature, but modularity is a key in its design. It would allow the use of different kinds of frontends, including those of a graphical nature.
Actually a GTK+ based graphical frontend is being actively developed ( see ["DebianInstallerGUI"] ) : this was built on the GTK frontend that has been developed during the last years by many members of the debian-installer team.
sarge will not ship with any kind of graphical installer; the project is simply way too unfinished. The main reason for this is that creating a graphical installer around DebianInstaller's current design is inherently very hard; just adding GTK+ or Qt widgets on top of cdebconf will not produce a friendlier installer than what is already available from the newt front end. In order to make a user-friendly graphical installer, one will have to do a lot of thinking as of how to combine multiple questions into one screen, do partitioning etc. most effectively in a graphical setting while still working well with the rest of the installer. This is definitely something that will not be done for sarge, but any help for future releases would be greatly appreciated.
Question 6: Will the DebianInstaller support creating and installing to software RAID devices?
Answer: The installer supports RAID 0, 1, and 5. Note that if you use RAID for /boot, you will be forced to use lilo since grub doesn't support RAID. ([http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=251905 #251905]).
[http://linux.yyz.us/sata/faq-sata-raid.html SATA RAID] Chipset's or PCI/PCIe-card's software RAIDs are not supported yet, or not completely, see dmraid [http://lists.debian.org/debian-boot/2005/11/msg00297.html 1] [http://packages.debian.org/unstable/admin/dmraid 2], if you do not need to dual boot to a MS Windows installed on a such software raid system, please use the debian-installer default raid manager, above mentioned, it is far more tested.
Question 7: How do I install using LVM? Is there any more information about using the LVM installer? LVM installation doesn't work!
Answer: LVM is fairly straightforward to configure using the menu item in the partitioner titled "Configure the Logical Volume Manager". Please note that it will probably help a lot if you know some [http://www.tldp.org/HOWTO/LVM-HOWTO/ LVM terminology] - it's recommended that you know the following three concepts: volume group (VG), physical volume (PV) and logical volume (LV) and how these relate to each other.
You can even build an LVM on top of a RAID device even for root. Nevertheless Grub has still some troubles ([http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=251905 #251905]).
Question 8: Is installation on SATA harddrives supported by DebianInstaller?
Answer: There are currently three options for SATA users:
- You can try the 2.4.27 kernel which is default in the sarge installer. This kernel includes some support for SATA.
- Install using the Linux 2.6 kernel which should have better support for your SATA hardware (boot the installer with "linux26"). A newer version of the 2.6 kernel than what was released in sarge is available in recent daily builds.
- See if you can change your SATA settings in the BIOS from something like "Native mode" to "Compatibility mode" (might be labeled differently)
For a list of supported chipsets and their status of support you may want to be assisted by [http://www.linuxmafia.com/faq/Hardware/sata.html]
If you have a Gigabyte GA-["K8NS"] Pro or similar mainboard, you may want to take a look at [http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=281843 #281843] (for i386 installation) or [http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=281839 #281839] (for ["AMD64"] installation).
Question 9: DebianInstaller does not include a driver for my xyz network card, but I have (tarball) driver sources. What are the steps for building that driver? Do the steps include the phrase, "Assume you have kernel-headers?" Do you think you'll be able to include kernel-headers for the installer's kernel image(s) in the CD image, so we don't have to sneaker-net them around?
Answer: It's possible to do this, but we do not yet have good end-user docs explaining how to do so. We're waiting for an end user to accomplish it and write them from experience. We're glad to help you through the process when you mail us at email@example.com (do not forget to add lspci -v output).
Question 10: Will there be support for Linux 2.6 kernels?
Answer: Boot with "linux26". Note that 2.6 is the default for at least powerpc installs, is not yet available for all architectures, and is not available when booting i386 from floppies. It is the default in daily builds for more architectures, including i386.
Question 11: How to install with boot floppies + netinst.iso mounted as a loop device in the ramdisk from a ext2 partition?
Answer: You can boot the installer directly from a hard disk using LILO or GRUB, and the hd-media initrd will find your netinst.iso. See the ["InstallationHOWTO"] for details
Question 12: How can I create boot iso with my own set of packages?
Answer: see ?DebianInstallerModify
Question 13: How can I build the DebianInstaller?
Answer: see ?DebianInstallerBuild
Question 14: Why after installer tells me it's finished and reboot it ejects the cdrom and then after the machine start up, it enters the welcome debian screen, but then it keep rebooting?
Answer: This [http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=249600&archive=yes bug #249600 against hotplug] should be fixed now. Please let us know if not.
Question 15: Why does the installer ask for the ide-scsi module to install the system, when it clearly can use the normal ide-cd module to do this? Why is the ide-scsi module not included on the cd install, and where can we get it (I have tried the cd drivers boot floppy, no good)?
Answer: Please file an installation report with the details if you see this; we never have encountered such a problem. The installer does not know about, or need the ide-scsi hack.
Question 16: Why do I have to choose twice which mirror to use (at installation and later when base-config is performed)?
Answer: This is a known bug and it's being worked on (see bug number [http://bugs.debian.org/220269 220269] for the current status).
Question 17: Why doesn't the installer automatically choose the appropriate kernel to install (for example an i686 kernel instead of the generic i386)?
Answer: The netinst CD image includes only the generic i386 kernel, so if you're using that you will always have install the proper kernel for your system manually after the installation.
If using any other installation method than the netinst CD you should get a kernel installed that is optimised for your hardware.
Question 18: What about SMP support?
Answer: The installer itself does not need any SMP support and will not boot with an SMP enabled kernel since it merely needs to install the system to your machine.
When the system is being installed, it would be nice if SMP was automatically detected and the appropriate kernel was chosen though (which would mean that SMP was enabled when you boot into your newly installed Debian system).
Question 19: Why isn't the installed system configured to use the language I chose during installation?
Answer: This problem has been fixed, so if you see it using any current image, please let us know.
Question 20: Why isn't X configured to use the keyboard I chose during installation?
Answer: This is a known problem, to check if it has been fixed or to track its progress, see bug report [http://bugs.debian.org/238778 238778].
Question 21: When I run the DebianInstaller, my keyboard doesn't work!
Answer: This seems to be especially common on laptops. Try booting with no APIC and no local APIC support ("boot: linux noapic nolapic").
Question 22: Why isn't my NIC supported by DebianInstaller, I know the "tg3" driver supports it!
Answer: As tg3 contains firmware which does not seem to meet the requirements of DFSG (the Debian Free Software Guidelines), a decision was made to remove the driver from debian packaged kernels (more information [http://lists.debian.org/debian-devel/2004/04/msg00405.html here] and [http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=239952 here]). Now this has mostly been resolved with a version of the driver that does not contain such firmware and should work with most cards. You should not have problems with the latest version of the installer. (On 9/14/2005 this was certainly not true. The Current daily of the netinst disc does not recognize the presence of a tg3 NIC when I tried to install.) (On 9/21/2005 I found a workaround for tg3 installs. The tg3 hardware is autodetected so the "Detect network hardware" dialog will fail. When it does, flip to the second virtual terminal and manually load the tg3 driver with "modprobe -a tg3" flip back to the first terminal and enter "No" to the ?FireWire ethernet question and then Go Back and restart the "Detect network hardware" dialog and everything should proceed ok from there.)
(FYI: last I heard the Debian kernel maintainers refused to maintain a forked tg3, and the upstream kernel maintainers refused to accept any changes. Currently the non-free version is in Debian, but it will probably have to be removed again, unnecessary though that is --Nathanael Nerode)
Question 23: There seem to be errors reading the CD!
Answer: First of all, check that the ["MD5"]-sum of the ISO file you downloaded is correct; the md5sum for each of the ["ISOs"] is available in the same directory as the ISO download. After booting the installer you can also check the CD-ROM: use the <go back> button to back up to the main menu and select the option to check the CD-ROM's integrity.
If you are positive that the CD is correct, it could be that the CD-ROM is old and/or flaky when using DMA. You can try disabling DMA for the CD-ROM (by booting with "linux ide=nodma"); unfortunately this boot option is currently broken (see bug [http://bugs.debian.org/226057 226057]).
You can also disable DMA using the following procedure.
- switch to VT2 (using alt-F2)
cd /proc/ide/hdX (X varies depending on where your CD drive is connected)
- check the current setting for using_dma
change it using the command echo "using_dma:0" >settings (or ":1")
- switch back to VT1 (using alt-F1)
A tightly folded IDE cable can also cause read errors, try repositioning the cable.
Question 24: How can I install sid (unstable) with DebianInstaller?
Answer: Start an installation in expert mode. Before||||After choosing the mirror you will be asked which distribution to install: stable, testing or unstable. We recommend using a daily build of the installer to install testing or unstable.
Question 25: How do I copy d-i (log)files to a remote host?
Answer 0: Use the included mini web server:
- This method requires a newer version of the installer than shipped with sarge, but it is the easiest way.
- Get to the main menu and choose the "Save debug logs" menu item.
- Tell it to start up the web server, and follow the prompts to download the logs from the web server to your other computer. If your other machine runs a version of debian after sarge, you can also use "apt-get install installation-report reportbug; reportbug installation-report" and follow the prompts to let it download the log files for you and produce a nice installation report.
Note: Both methods described below will only work after you have configured a network interface, and assume some familiarity with the shell.
Answer 1: Use netcat (included in busybox):
- Make sure you have netcat installed on the remote system.
On the remote system: start a shell, cd to a directory where you wish to save the files, and run nc -l -p 1234 >syslog. This will make the remote system listen on port 1234 for a connection from another netcat process.
On the system you are installing: go to ["VT2"] and run 'nc <remote_system> 1234 <remote_system>' by the name or ip address of the remote system.
Answer 2: Use the open-ssh-client installer module:
- Start the installer in expert mode (or medium priority).
At Load installer components select open-ssh-client-udeb from the list.
From the system you are installing, you should now be able to use scp to copy files to a remote system using e.g. 'scp /var/log/syslog <username>@<remote_system>:', again replacing <remote_system> by the name or ip address of the remote system.
By default the file will be copied to the home directory of <username> on the remote system; optionally a path can be added after the ':'.
In the examples above I've used /var/log/syslog as the file to be transferred and, for netcat, 1234 as the port to be used for the transfer; both can of course be changed.
Both methods above can also be used to copy files from a remote system to the system being installed. This is left as an exercise to the reader.
Question 26: How do I use netinst over ppp(oe)?
You can't use PPP or ["PPPoE"] with the businesscard CD-ROM image! That's because these packages are not included in it. With the netinst CD-ROM image this is done in 2nd stage (= after first reboot).
PPP configuration is provided automatically by pppconfig, if you skipped Configure the network in 1st stage (= before first reboot). If you accidently configure network in 1st stage, then you have to choose Configure and enable ppp manually in the 2nd stage main menu. If you are not using expert mode, then you have to press Cancel when Debian-Installer gets to Apt configuration.
Be aware that so called winmodems are not supported yet!
Network configuration is not automatic for machines using ["PPPoE"] for their ADSL connection. It it possible to setup an ADSL/["PPPoE"] connection with the Debian-Installer but there isn't any menu item for it in the installer yet. This will be addressed soon. In the meanwhile the workaround is, at the beginning of the second phase (after the reboot done by the installer), to switch to a virtual console (Ctrl-Alt-Functionkey) and then run pppoeconf from here.
Network configuration fails for computers using router modems like Alcatel ?SpeedTouch 530, that is some DSL modems which provide both DHCP and PPP services. Using Debian-Installer with such a configuration is still possible but not automated yet. The manual workaround is described in the [http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=265183 #265183] bug report.
Question 27: What do I do if I can no longer boot Windows after installing Debian? If all goes well, Debian Installer should automatically add a menu item for an existing Windows installation. Currently there are two known issues that may result you not being able to boot into Windows after the installation.
Issue 1: There is no menu item in the bootloader If you decided to select a mountpoint for your Windows partition during partitioning, the installer may have been unable to find Windows due to bug [http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=251794 #251794]. You can fix this by manually adding the Windows partition to your bootloader configuration.
If you are using GRUB as your bootloader, add the following lines at the end of the file /boot/grub/menu.lst:
title Windows root (hd0,0) savedefault makeactive chainloader +1
You should change (hd0,0) to refer to the correct partition: (hd0,0) is Grub's notation for the first partition on the first harddisk (/dev/hda1 or /dev/sda1); (hd1,3) is the fourth partition on the second harddisk (/dev/hdb4).
If you are using LILO as your bootloader, add the following lines at the end of the file /etc/lilo.conf:
Again, change /dev/hda1 to refer to your Windows partition. After updating this file, you will need to run 'lilo' as root.
Issue 2: There is a menu entry, but you get errors when you try to boot Windows There could be different reasons for this. First you should check that the configuration of your bootloader is correct. See issue 1 for some pointers.
If you installed using the 2.6 kernel, another cause may be that the hard disk geometry as reported in the partition table was altered during installation. See [http://lwn.net/Articles/86835 this article] in LWN for a good explanation of this issue and information on how to rectify it.
Question 28: Can I use d-i as a rescue system? What pitfalls are there?
Yes, you can use d-i to mount your hard disks and fix /etc/shadow if you forgot your password.
And you can also use d-i to boot your system to chroot into your installation and execute your boot-loader (grub, lilo, yaboot, ...) if you cannot boot from hard disk any more.
If you have a version of the installer that is later than that released in sarge, proper rescue support is included, just boot the installer with "rescue" and follow the prompts until it mounts your hard disk and gives you a shell chrooted into your system.
Otherwise, just boot the installation system as if you were starting a new installation, proceed to the point where you get to partitioning harddisks and stop there. Next switch to ["VT2"] (using alt-F2) or back out to the menu and start a shell from there. Create a directory /target, manually mount the partition(s) where you installed Debian on /target and execute chroot /target. Note: the installer uses devfs, so you will find your harddisks under /dev/discs. You cannot use the d-i menu item "install boot-loader" because the menu systems uses dependencies which require a Debian install (on a clean partition...) which you don't want.
Rescue with post-Sarge daily built images
The post-Sarge daily built images have a new rescue boot option. This will use the first part of the installation process to set up a working system, ask you to select the device that holds your root partition, and will then mount that partition and chroot to it.
Note: you will need to mount any other file systems (including proc) manually, despite what 'mount' will tell you.
Question 29: How to use preseeding? Howto use network console with preseeding?
Question 30: How to shrink an existing windows installation?
Answer: please see the manual, chapter 6.3.2. Basically, you need to select the item displaying the size of the partition to change the size. Please make sure that you check the partition for correctness before (best with chkdsk), and that you have a backup of your data, and also are able to re-install Windows (or forget about it totally) in case something breaks.
List of changes
- FJP -- 2005-Okt-13
- Document how to disable DMA for CD-ROM drives manually
- FJP -- 2005-Aug-15
- Document rescue option for post-Sarge images
- JoeyH -- 2005-Aug-3
- Many improvements and updates.
- ["ABa"] -- 2005-May-27
- Windows shrinking
- ["LiamOB"] -- 2005-May-19
- Noted location of ["MD5sums"] for ["ISOs"]
- h01ger -- 2005-Feb-22
- finally answered #29
- ["LMClapp68"] -- 2005-Jan-29
- s/its/it's/ where appropriate, and vice versa
- spell checked
- h01ger -- 2005-Jan-25
- corrected answer 24 (it is now possible to install unstable with d-i
- added Question 29
- FJP -- 2004-Dec-19
- Added extra instructions on using d-i as rescue system
- Improved some formatting
- h01ger -- 2004-Dec-14
- added entry about rescue system
- h01ger -- 2004-Dec-13
- Answered question 26 (copied from debian-boot) on the pppoe part
- ATLF -- 2004-Dec-09
- Changed "hd(0,0)" to "(hd0,0)"
- FJP -- 2004-Dec-06
- Added question 27 on Windows boot problems
- Added info on Gigabyte GA-["K8NS"] Pro mainboards to question 8
Mars ?DeLapp -- 2004-Nov-02
- Added info on CD read errors
- FJP -- 2004-Oct-31
- reverted to version 1.99, which is the last properly formatted version
- FJP -- 2004-Oct-06
- Modified info on copying files to a remote system
- Sesse -- 2004-Aug-27
- Added more information on why we don't have a graphical installer in sarge
?TheOrb -- 2004-Aug-10
- Added link to bug and more info on how to work around it to question 14.
- FJP -- 2004-July-12
- Added info on copying files using ssh
- Osamu -- 2004-June-18
- Added info on DHCP/FIX IP issue
- AlexeyS -- 2004-June-06
- Added question 26 about ru_RU
- DavidH -- 2004-05-09
- Answered sid question
- DavidH -- 2004-05-06
- Added CD-ROM read errors question
- DavidH -- 2004-05-05
- Added X keyboard question
- Added dead keyboard question
- Added missing tg3 driver question
- Corrected question numbering
- DavidH -- 2004-05-03
- Added more information to LVM question
- Added 3 more questions (locales, SMP, auto-kernel selection)
- Changed layout a bit, added more text to initial Note