[SLE] General question about the rescue system
Used to be that if you wanted to do major things to your normal system while it was 'broken', you could: 1) boot the rescue system from CD or DVD 1 2) mount your root partition on /mnt 3) chroot /mnt 4) Mount your boot partition on /boot and there you would for most purposes have your running system. You could change your fstab or re-install grub or whatever that might be needed. However, with 10.1 and the udev system, the devices are no longer available to you after the chroot. If you do the first 3 steps above, there is no way to do step 4 because there are no hard drives 'seen' by the system. Their definitions are all left behind in the /dev/ of the rescue system. Is there a way around this? One could copy the /dev from the rescue system to the normal system but I am not sure that would work or whether it would be a good idea. A quick look in the starter manual shows no reference to the rescue system. On a 'lab rat', I just did copy the /dev/* from the rescue system over to the normal system (savi9ng the normal /dev first) and it *does* work. But there must be a better way. -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
* Bruce Marshall
Is there a way around this? One could copy the /dev from the rescue system to the normal system but I am not sure that would work or whether it would be a good idea. A quick look in the starter manual shows no reference to the rescue system.
On a 'lab rat', I just did copy the /dev/* from the rescue system over to the normal system (savi9ng the normal /dev first) and it *does* work.
But there must be a better way.
What about booting with the install disk, choosing install and when you get to the screen with "Install/Boot from HardDrive" choose the "Other" box and repair installed system? AIR, there is an option for fixing the partition table and another for the boot process. -- Patrick Shanahan Registered Linux User #207535 http://wahoo.no-ip.org @ http://counter.li.org HOG # US1244711 Photo Album: http://wahoo.no-ip.org/gallery2 -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On Friday 21 July 2006 15:22, Patrick Shanahan wrote:
But there must be a better way.
What about booting with the install disk, choosing install and when you get to the screen with "Install/Boot from HardDrive" choose the "Other" box and repair installed system? AIR, there is an option for fixing the partition table and another for the boot process.
Well, there are lots of things that YAST could probably do for you but I'm thinking that the way it used to work was best. I ran into the problem recently when I replaced the hard drives in my wife's machine and needed to re-install grub with some new definitions.... The strange thing was, that when running the rescue system, it saw the new drives (SATA) as SDB and SDA whereas booting from the drives, the system saw them as the reverse of that. So I couldn't really install grub from the rescue system. It was a weird problem. -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On Friday 21 July 2006 20:27, Bruce Marshall wrote:
Used to be that if you wanted to do major things to your normal system while it was 'broken', you could:
1) boot the rescue system from CD or DVD 1
2) mount your root partition on /mnt
3) chroot /mnt
4) Mount your boot partition on /boot
and there you would for most purposes have your running system. You could change your fstab or re-install grub or whatever that might be needed.
However, with 10.1 and the udev system, the devices are no longer available to you after the chroot. If you do the first 3 steps above, there is no way to do step 4 because there are no hard drives 'seen' by the system. Their definitions are all left behind in the /dev/ of the rescue system.
Is there a way around this? One could copy the /dev from the rescue system to the normal system but I am not sure that would work or whether it would be a good idea. A quick look in the starter manual shows no reference to the rescue system.
On a 'lab rat', I just did copy the /dev/* from the rescue system over to the normal system (savi9ng the normal /dev first) and it *does* work.
But there must be a better way.
Standard procedure is mount -o bind /dev /mnt/dev mount -o bind /proc /mnt/proc mount -o bind /sys /mnt/sys before you do the chroot -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On Friday 21 July 2006 15:24, Anders Johansson wrote:
But there must be a better way.
Standard procedure is
mount -o bind /dev /mnt/dev mount -o bind /proc /mnt/proc mount -o bind /sys /mnt/sys
before you do the chroot
Sounds like a plan. If this does the trick, I'm going to have it tatoo'd in a visible spot on my body... <grin> -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On Friday 21 July 2006 16:13, Bruce Marshall wrote:
Standard procedure is
mount -o bind /dev /mnt/dev mount -o bind /proc /mnt/proc mount -o bind /sys /mnt/sys
before you do the chroot
Sounds like a plan. If this does the trick, I'm going to have it tatoo'd in a visible spot on my body... <grin>
And it works. Thanks! -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On 21/07/06 14:13, Bruce Marshall wrote:
On Friday 21 July 2006 15:24, Anders Johansson wrote:
But there must be a better way.
Standard procedure is
mount -o bind /dev /mnt/dev mount -o bind /proc /mnt/proc mount -o bind /sys /mnt/sys
before you do the chroot
Sounds like a plan. If this does the trick, I'm going to have it tatoo'd in a visible spot on my body... <grin>
Better not have it tatooed, because surely "/mnt" will someday be changed to something else.. and then you'll be lost ;-) The manpage for mount has left me a little confused (as manpages so often do -- never let the programmer write the documentation :-) ). Suppose I've done something like this in my fstab: /dev/hdb1 /var (etc) /dev/hdb2 /var/log (etc) and now I want to mount the entire /var tree under /mnt/var, so the logs will be available as /mnt/var/log. It seems to me that mount -o bind /var /mnt/var will not do the trick (/var/log would not be remounted as /mnt/var/log), rather I should use mount -o rbind /var /mnt/var Yes/no? -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On Saturday 22 July 2006 00:06, Darryl Gregorash wrote:
On 21/07/06 14:13, Bruce Marshall wrote:
On Friday 21 July 2006 15:24, Anders Johansson wrote:
But there must be a better way.
Standard procedure is
mount -o bind /dev /mnt/dev mount -o bind /proc /mnt/proc mount -o bind /sys /mnt/sys
before you do the chroot
Sounds like a plan. If this does the trick, I'm going to have it tatoo'd in a visible spot on my body... <grin>
As a Memento? :)
Better not have it tatooed, because surely "/mnt" will someday be changed to something else.. and then you'll be lost ;-)
The manpage for mount has left me a little confused (as manpages so often do -- never let the programmer write the documentation :-) ). Suppose I've done something like this in my fstab:
/dev/hdb1 /var (etc) /dev/hdb2 /var/log (etc)
and now I want to mount the entire /var tree under /mnt/var, so the logs will be available as /mnt/var/log. It seems to me that
mount -o bind /var /mnt/var
will not do the trick (/var/log would not be remounted as /mnt/var/log), rather I should use
mount -o rbind /var /mnt/var
Yes/no?
Apparently so, yes But you can't use rbind on the root partition in Bruce's problem, since the root wasn't mounted anywhere else to begin with, if that was what you were aiming for -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On 21/07/06 17:23, Anders Johansson wrote:
But you can't use rbind on the root partition in Bruce's problem, since the root wasn't mounted anywhere else to begin with, if that was what you were aiming for
No, I saw that it wouldn't apply there.. it wouldn't make sense to mount those anyway in that problem, until after the chroot. My question should have gone as a new thread, or at the very least with a modified subject, which I realized after I sent the mail. -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Friday 2006-07-21 at 14:27 -0400, Bruce Marshall wrote:
However, with 10.1 and the udev system, the devices are no longer available to you after the chroot. If you do the first 3 steps above, there is no way to do step 4 because there are no hard drives 'seen' by the system. Their definitions are all left behind in the /dev/ of the rescue system.
It is possible to install a fixed dev tree (package devs). This gets superseded as soon as udev runs, but if it fails, you have the other, fixed, one. I thought every body had this :-? - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFEwS7XtTMYHG2NR9URAosiAJ4z4cEVVbmcYt6eAIOGChv8JxCTPQCfXwBx u/wdaSYfuBGSLzsMlnkmZZ0= =VIAE -----END PGP SIGNATURE----- -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
On Friday 21 July 2006 15:45, Carlos E. R. wrote:
The Friday 2006-07-21 at 14:27 -0400, Bruce Marshall wrote:
However, with 10.1 and the udev system, the devices are no longer available to you after the chroot. If you do the first 3 steps above, there is no way to do step 4 because there are no hard drives 'seen' by the system. Their definitions are all left behind in the /dev/ of the rescue system.
It is possible to install a fixed dev tree (package devs). This gets superseded as soon as udev runs, but if it fails, you have the other, fixed, one. I thought every body had this :-?
I have devs on this system but have no idea how to use it. And the 'lab rat' that I mentioned that I copied the rescue system /dev to seems to be permanently fixed. Even though I tried to replace the original /dev directory to restore it to original status, it still retains the nodes for the hard drives. -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Friday 2006-07-21 at 16:16 -0400, Bruce Marshall wrote:
It is possible to install a fixed dev tree (package devs). This gets superseded as soon as udev runs, but if it fails, you have the other, fixed, one. I thought every body had this :-?
I have devs on this system but have no idea how to use it.
It can not be installed on a live system when udev is running, I think. The behaviour is similar to "mount": if you mount something into a directory that already has contents, these disappear, become unavailable, even if they are there somewhere. Only the mounted things are visible. When you umount, the old directory contents are visible again. Dunno if I explained myself well, I'm sleepy. Thus the devs package acts as a backup, only vissible when udevs is not running. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFEwWgqtTMYHG2NR9URAoY+AJ4yNYXSqQed0EdbOC2g8qmDTvJCOwCfXf6f Qui9EaPVMMOCAuKOKwn1APo= =95Wj -----END PGP SIGNATURE----- -- Check the headers for your unsubscription address For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com Please read the FAQs: suse-linux-e-faq@suse.com
participants (5)
-
Anders Johansson
-
Bruce Marshall
-
Carlos E. R.
-
Darryl Gregorash
-
Patrick Shanahan