Menu
▾
▴
user-mode-linux-user
|
From: Paul W. <pd...@ex...> - 2004-05-01 19:41:19
|
I found that instructions for installing Debian under UML were a bit thin on the ground, so I wrote some: http://www.ex-parrot.com/~pdw/debian-uml-install.html I hope you find them useful. cheers, Paul |
|
From: Jody B. <lis...@pi...> - 2004-05-01 20:05:01
|
On Sat, May 01, 2004 at 08:41:17PM +0100, Paul Warren wrote: > I found that instructions for installing Debian under UML were a bit > thin on the ground, so I wrote some: > > http://www.ex-parrot.com/~pdw/debian-uml-install.html I put this together recently. what do you think? http://uml.pimb.org/howto/#uml-debian-install -- Jody Belka knew (at) pimb (dot) org |
|
From: henrique <hen...@fu...> - 2004-05-01 20:38:38
|
Great stuff! Currently I have one with ~400 Mb with a lot of pkgs, and I am downsizing it for suiting a 'busybox'-based fs. I also find very interesting the following commands to cleanup large root file-systems prepared. Example: do not care for other dependencies: => dpkg --force-all --auto-deconfigure --remove tk8.4 A very useful hint for installing: UML running inside intranet, access through SQUID on 192.168.0.1, add line in /etc/apt/apt.conf: => Acquire::http::Proxy "http://192.168.0.1:3128"; and ignore cache limits: => APT::Cache-Limit 10000000; And another for selecting 'testing' pkgs, at /etc/apt/sources.list: => deb http://non-us.debian.org/debian-non-US testing/non-US main contrib non-free => deb http://ftp.uk.debian.org/debian/ testing main non-free contrib ...and expanding the 'dpkg -l' command... COLUMNS=100 dpkg -l For UML, an interesting way of avoiding HW access to clock, thus some anoying warnings: /etc/default/rcS: # Do not use hardware clock in UML to avoid small warnings HWCLOCKACCESS=no <==<== Good woody for all! > > I found that instructions for installing Debian under UML were a bit > > thin on the ground, so I wrote some: > > > > http://www.ex-parrot.com/~pdw/debian-uml-install.html > > I put this together recently. what do you think? > > http://uml.pimb.org/howto/#uml-debian-install |
|
From: Paul W. <pd...@ex...> - 2004-05-01 22:24:39
|
On Sat, May 01, 2004 at 08:06:03PM +0000, Jody Belka wrote: > I put this together recently. what do you think? > > http://uml.pimb.org/howto/#uml-debian-install Looks good. A few queries: - Is there a reason you make the ext2 filesystems prior to running the installer? - Did you not hit the problem with the installer not being able to find any kernel modules? I prefer to use the fakeide/fakehd stuff because other distributions' installers are not so clued up when it comes to ubd devices, and using fakeide/fakehd universally makes admin easier. Jeff - any chance of a link to these from: http://user-mode-linux.sourceforge.net/install.html ? cheers, Paul |
|
From: Jody B. <lis...@pi...> - 2004-05-01 22:48:09
|
On Sat, May 01, 2004 at 11:24:26PM +0100, Paul Warren wrote: > - Is there a reason you make the ext2 filesystems prior to running the > installer? Well, it was actually ext3. But the reason why is that if you wanted to use a filesystem type for which there weren't any tools in the installer image you would have to create the filesystem outside, so I thought i'd do it that way even with ext2/3. I'll add a note about this to the HOWTO. > - Did you not hit the problem with the installer not being able to find > any kernel modules? No, because the *only* bits i used in the installer were keyboard, hostname, networking and base system. That's all you need to run. There's no reason to run the "configure device drivers stage". -- Jody Belka knew (at) pimb (dot) org |
|
From: Paul W. <pd...@ex...> - 2004-05-02 00:15:08
|
On Sat, May 01, 2004 at 10:50:15PM +0000, Jody Belka wrote: > > - Did you not hit the problem with the installer not being able to find > > any kernel modules? > > No, because the *only* bits i used in the installer were keyboard, > hostname, networking and base system. That's all you need to run. There's > no reason to run the "configure device drivers stage". OK, this is because I'm not that familiar with the Debian installer - it didn't occur to me that I could get past that stage without resolving the problem. cheers, Paul |
|
From: Millis M. <mi...@fa...> - 2004-05-02 02:39:56
|
After following the instructions, and starting a network install, I get errors saying (in order): Unable to mount the Rescue Floppy. The attempt to extract the Rescue Floppy failed. Installation of the file '/tmp/rescue.bin' failed. and I am prompted again to start the install. Any ideas? (Otherwise, the document was clear and easy to follow). Thanks, Millis On Sat, 2004-05-01 at 21:06, Jody Belka wrote: > I put this together recently. what do you think? >=20 > http://uml.pimb.org/howto/#uml-debian-install |
|
From: Paul W. <pd...@ex...> - 2004-05-02 10:58:39
|
On Sun, May 02, 2004 at 03:39:46AM +0100, Millis Miller wrote: > After following the instructions, and starting a network install, I get > errors saying (in order): > > Unable to mount the Rescue Floppy. > The attempt to extract the Rescue Floppy failed. > Installation of the file '/tmp/rescue.bin' failed. Have you got MSDOS file system and loopback filesystem support compiled into your kernel? Otherwise, are there any log messages on the other consoles? Paul |
|
From: Millis M. <mi...@fa...> - 2004-05-02 14:39:19
|
This seems to be the problem: I am using the Debian UML kernel (2.4.24-1um) which has loopback compiled in but MSDOS is only a module, and that's not obviously in the installer image. Am working out how to get around this, thanks for the tip. Millis On Sun, 2004-05-02 at 11:58, Paul Warren wrote: > On Sun, May 02, 2004 at 03:39:46AM +0100, Millis Miller wrote: > > After following the instructions, and starting a network install, I get > > errors saying (in order): > >=20 > > Unable to mount the Rescue Floppy. > > The attempt to extract the Rescue Floppy failed. > > Installation of the file '/tmp/rescue.bin' failed. >=20 > Have you got MSDOS file system and loopback filesystem support compiled > into your kernel? >=20 > Otherwise, are there any log messages on the other consoles? >=20 > Paul >=20 >=20 > ------------------------------------------------------- > This SF.Net email is sponsored by: Oracle 10g > Get certified on the hottest thing ever to hit the market... Oracle 10g.=20 > Take an Oracle 10g class now, and we'll give you the exam FREE.=20 > http://ads.osdn.com/?ad_id=3D3149&alloc_id=3D8166&op=3Dclick > _______________________________________________ > User-mode-linux-user mailing list > Use...@li... > https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user >=20 |
|
From: Paul W. <pd...@ex...> - 2004-05-02 15:04:27
|
On Sun, May 02, 2004 at 03:39:12PM +0100, Millis Miller wrote: > This seems to be the problem: I am using the Debian UML kernel > (2.4.24-1um) which has loopback compiled in but MSDOS is only a module, > and that's not obviously in the installer image. > > Am working out how to get around this, thanks for the tip. You could try my kernel image: http://www.ex-parrot.com/~pdw/linux-2.4.26-20040429.gz Paul |
|
From: David C. <li...@ed...> - 2004-05-01 22:31:52
|
On Saturday 01 May 2004 20:41, Paul Warren wrote: > I found that instructions for installing Debian under UML were a bit > thin on the ground, so I wrote some: > > http://www.ex-parrot.com/~pdw/debian-uml-install.html I use debootstrap to create Debian filesystems, I haven't tried booting one under UML yet but they work fine for TFTP boots over a network on real machines. Nice tutorial though and thanks for the link to a newer initrd, I might try booting that off of the network using pxelinux. David |
|
From: Mike S. <msn...@ln...> - 2004-05-06 02:30:35
|
On Sat, May 01 2004 at 16:31, David Cannings <li...@ed...> wrote: > On Saturday 01 May 2004 20:41, Paul Warren wrote: > > I found that instructions for installing Debian under UML were a bit > > thin on the ground, so I wrote some: > > > > http://www.ex-parrot.com/~pdw/debian-uml-install.html > > I use debootstrap to create Debian filesystems, I haven't tried booting > one under UML yet but they work fine for TFTP boots over a network on > real machines. Also there is debian's rootstrap; rootstrap runs debootstrap within a UML instance. I'm using it to automate the creation of custom Debian images with great success. Mike |
|
From: Shed. <sh...@be...> - 2004-05-06 22:37:40
|
Mike, Could you share some *examples* of how you are using rootstrap? Mike Snitzer wrote: > > Also there is debian's rootstrap; rootstrap runs debootstrap within a UML > instance. I'm using it to automate the creation of custom Debian images > with great success. > > Mike > |
|
From: Peter <pe...@ri...> - 2004-05-04 11:32:41
|
Hi Paul. Thanks for the howto. It was definitely a 'cool, I didn't = realise you could do that moment' for me. I tried this howto out on a WhiteBox Linux http://whiteboxlinux.org = install (basically a Red Hat Advanced Server 3 distro build from = sources) to see if the same concept would work there. Since if we can = just use a regular installer to get UML file systems created, then the = creation of UML file systems becomes trivial and it'd be much easier to = get a UML running your personal preference distro. The installer from = ftp://mirror.physics.ncsu.edu/pub/whitebox/3.0/en/os/i386/images/ boots = up just fine. It progresses through the install, hard disk = partitioning, package selection, even package installation. Then the UML = instance reboots (and brings back up the installer). I kill the UML = linux process. But after the installer has done its thing I have a 'what next = question'. How do I activate the next phase of the install process? The root files = system does not seem able to boot up. Nor be readable by the any of the = usual methods I use. e.g. I get this when using it to boot up a UML = instance: kjournald starting. Commit interval 5 seconds EXT3-fs: mounted filesystem with ordered data mode. VFS: Mounted root (ext3 filesystem) readonly. Warning: unable to open an initial console. Kernel panic: No init found. Try passing init=3D option to kernel. I try to mount the root file system on the host but get:=20 # mount -o loop -t ext3 rootfs.ext3 tmp/ mount: wrong fs type, bad option, bad superblock on /dev/loop1, or too many mounted file systems Does anyone have any suggestion how I can read the rootfs created by the = install program?=20 Cheers, Peter ----- Original Message -----=20 From: "Paul Warren" <pd...@ex...> To: <Use...@li...> Sent: Sunday, May 02, 2004 7:41 AM Subject: [uml-user] Debian install instructions > I found that instructions for installing Debian under UML were a bit > thin on the ground, so I wrote some: >=20 > http://www.ex-parrot.com/~pdw/debian-uml-install.html >=20 > I hope you find them useful. >=20 > cheers, >=20 > Paul >=20 >=20 >=20 > ------------------------------------------------------- > This SF.Net email is sponsored by: Oracle 10g > Get certified on the hottest thing ever to hit the market... Oracle = 10g.=20 > Take an Oracle 10g class now, and we'll give you the exam FREE.=20 > http://ads.osdn.com/?ad_id=3D3149&alloc_id=3D8166&op=3Dclick > _______________________________________________ > User-mode-linux-user mailing list > Use...@li... > https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user >=20 > |
|
From: Paul W. <pd...@ex...> - 2004-05-04 11:50:20
|
On Tue, May 04, 2004 at 11:30:25PM +1200, Peter wrote: > I tried this howto out on a WhiteBox Linux http://whiteboxlinux.org > install (basically a Red Hat Advanced Server 3 distro build from > sources) to see if the same concept would work there. Since if we can > just use a regular installer to get UML file systems created, then the > creation of UML file systems becomes trivial and it'd be much easier > to get a UML running your personal preference distro. > > The installer from > ftp://mirror.physics.ncsu.edu/pub/whitebox/3.0/en/os/i386/images/ > boots up just fine. It progresses through the install, hard disk > partitioning, package selection, even package installation. Then the > UML instance reboots (and brings back up the installer). I kill the > UML linux process. > > But after the installer has done its thing I have a 'what next question'. > > How do I activate the next phase of the install process? The root > files system does not seem able to boot up. Nor be readable by the > any of the usual methods I use. e.g. I get this when using it to boot > up a UML instance: kjournald starting. Commit interval 5 seconds > EXT3-fs: mounted filesystem with ordered data mode. VFS: Mounted root > (ext3 filesystem) readonly. Warning: unable to open an initial > console. Kernel panic: No init found. Try passing init= option to > kernel. Interesting - my next job is to do this for White Box, so your mail is very timely. What happens if you give it init=/bin/sh ? > I try to mount the root file system on the host but get: > # mount -o loop -t ext3 rootfs.ext3 tmp/ > mount: wrong fs type, bad option, bad superblock on /dev/loop1, > or too many mounted file systems I don't understand that bit fully - is it because there is now a partition table at the beginning of that disk? I suspect that there may be a second stage of the installer that also uses an initrd to get started. To make it work, you'd need to find that initrd, extract it, and pass it as an initrd option. Paul |
|
From: Paul W. <pd...@ex...> - 2004-05-04 16:04:42
|
On Tue, May 04, 2004 at 12:49:45PM +0100, Paul Warren wrote:
> > mount: wrong fs type, bad option, bad superblock on /dev/loop1,
> > or too many mounted file systems
>
> I don't understand that bit fully - is it because there is now a
> partition table at the beginning of that disk?
Yes it is. The way around this appears to be as follows -
First run fdisk on your image:
fdisk -l fedora-core-1-20040429
You must set heads sectors and cylinders.
You can do this from the extra functions menu.
Disk fedora-core-1-20040429: 0 heads, 0 sectors, 0 cylinders
Units = cylinders of 1 * 512 bytes
Device Boot Start End Blocks Id System
fedora-core-1-20040429p1 * 33 6291456 3145712 83 Linux
The important numbers seem to be 33 and 512. Take (33-1) * 512 = 16384,
and use that as an offset to the loopback mount:
mount fedora-core-1-20040429 tmp -o loop,offset=16384
I'm guessing that the "-1" is because the units are 1 based or
something?
I've been doing this by trial and error, but will it always be the case
that the partition table is 16k long?
Paul
|
|
From: Peter <pe...@ri...> - 2004-05-05 03:00:51
|
Paul. Thanks for the heads up on the 'offset' mount option. That was = just what I needed to know. Here is a precis of my install process. I = hope it works for you. I suspect the initrd UML boot method is robust = enough to install most any Linux distro (with the hypothesis being = tested on Debian and Whitebox Linux to date). # grab the boodisk image, and extract the initrd.img file wget = ftp://mirror.physics.ncsu.edu/pub/whitebox/3.0/en/os/i386/images/bootdisk= .img mkdir bootdisk mount bootdisk.img bootdisk # create a non-formatted file system. The installer will go and = partition this. dd if=3D/dev/zero bs=3D1k count=3D1 seek=3D$[4*1024*1024] of=3Drootfs ./linux mem=3D256M fakehd fake_ide ubd0=3Drootfs con0=3Dfd:0,fd:1 = root=3D6201 ubd=3D3 \ ramdisk_size=3D16384 con1=3Dpts con2=3Dpts con3=3Dpts con4=3Dpts = con5=3Dpts eth0=3Dtuntap,tap7 initrd=3Dbootdisk/initrd.img Note, I'm using tuntap for my eth0. You use whatever you use. You should be in the installer now... Set your IP info. Use this info for the FTP location of the RPMs: mirror.physics.ncsu.edu /pub/whitebox/3.0/en/os/i386/ Go with the default partitioning. I tried with just a single / = partition and a few other combinations. But I ran into problems when = the installer got to the meaty bit of actually installing rpms. Selected lilo boot loader (you can skip but I believe the rpms then = won't install after you select them). It should run through downloading/installing rpms. Happens rather = quickly on my 100Mbps connection. It will then reboot into the installer (again). Kill the UML instance. Figure out the block offsets per "sfdisk -l -uB rootfs". e.g. with = 16*1024 gets the /boot partition, and 102400*1024 gets the / partition. Device Boot Start End #blocks Id System rootfs.ext3p1 * 16 102399 102384 83 Linux rootfs.ext3p2 102400 3670015 3567616 83 Linux rootfs.ext3p3 3670016 4194303 524288 82 Linux swap rootfs.ext3p4 0 - 0 0 Empty # mount the / partition mount -t ext3 -o loop,offset=3D104857600 rootfs.ext3 tmp/ # create a 'regular' ubd device for your 'proper' file system dd if=3D/dev/zero bs=3D1k count=3D1 seek=3D$[4*1024*1024] of=3Dwbl mkfs.ext3 wbl mkdir tmp2 mount -t ext3 -o loop wbl tmp2 Then copy the files/directories from the root partition from the rootfs = into the wbl file system. I have a script for that. If people need it, = let me know. I'm excluding it here to keep this bit concise and on = topic. Then edit the things you'd normally do on the created file system: = tmp2/etc/fstab, tmp2/etc/inittab, etc. I'm now in my WBL UML instance and setting things up. It looks like its = working fine (i.e. I can ssh into it). - Peter ----- Original Message -----=20 From: "Paul Warren" <pd...@ex...> To: "Peter" <pe...@ri...> Cc: <Use...@li...> Sent: Wednesday, May 05, 2004 4:04 AM Subject: Re: [uml-user] Re: [insert distro name here] install = instructions > On Tue, May 04, 2004 at 12:49:45PM +0100, Paul Warren wrote: > > > mount: wrong fs type, bad option, bad superblock on /dev/loop1, > > > or too many mounted file systems > >=20 > > I don't understand that bit fully - is it because there is now a > > partition table at the beginning of that disk? >=20 > Yes it is. The way around this appears to be as follows - >=20 > First run fdisk on your image: >=20 > fdisk -l fedora-core-1-20040429 > You must set heads sectors and cylinders. > You can do this from the extra functions menu. > =20 > Disk fedora-core-1-20040429: 0 heads, 0 sectors, 0 cylinders > Units =3D cylinders of 1 * 512 bytes > =20 > Device Boot Start End Blocks Id = System > fedora-core-1-20040429p1 * 33 6291456 3145712 83 Linux >=20 > The important numbers seem to be 33 and 512. Take (33-1) * 512 =3D = 16384, > and use that as an offset to the loopback mount: >=20 > mount fedora-core-1-20040429 tmp -o loop,offset=3D16384 >=20 > I'm guessing that the "-1" is because the units are 1 based or > something? >=20 > I've been doing this by trial and error, but will it always be the = case > that the partition table is 16k long? >=20 > Paul >=20 >=20 >=20 > ------------------------------------------------------- > This SF.Net email is sponsored by: Oracle 10g > Get certified on the hottest thing ever to hit the market... Oracle = 10g.=20 > Take an Oracle 10g class now, and we'll give you the exam FREE.=20 > http://ads.osdn.com/?ad_id=3D3149&alloc_id=3D8166&op=3Dclick > _______________________________________________ > User-mode-linux-user mailing list > Use...@li... > https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user >=20 > |
|
From: Paul W. <pd...@ex...> - 2004-05-04 19:42:47
|
On Tue, May 04, 2004 at 11:30:25PM +1200, Peter wrote: > The installer from > ftp://mirror.physics.ncsu.edu/pub/whitebox/3.0/en/os/i386/images/ > boots up just fine. It progresses through the install, hard disk > partitioning, package selection, even package installation. You had more luck than me, then. I've just tried this and the installer claims I don't have any hard disks. Can you send me details of the command line and kernel that you used here? thanks, Paul |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X