What should I do to make an existing CentOS (5.4) disc boot up on a new computer? [...] Would it be enough to boot with a DVD in rescue mode, or boot with another hd, and install grub?
On 16.12.2009 12:16, Sorin Srbu wrote:
For me it has worked to just install the old hd in the new machine and boot it up. Kudzo takes care of the rest.
Then you have been lucky. :-) For me, the startup stopped already before the CentOS splash screen. I guess something was wrong with the initrd.
- Jussi
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On
Behalf
Of Jussi Hirvi Sent: Wednesday, December 16, 2009 11:39 AM To: CentOS mailing list Subject: Re: [CentOS] Old hd, new machine
On 16.12.2009 12:16, Sorin Srbu wrote:
For me it has worked to just install the old hd in the new machine and
boot
it up. Kudzo takes care of the rest.
Then you have been lucky. :-) For me, the startup stopped already before the CentOS splash screen. I guess something was wrong with the initrd.
Maybe, I always use a legacy custom setup for the partitions with a /boot, swap, / and a /home, install and set it up as I want it, and then clone and deploy the image. Has worked so far.
Things are simpler that way IMO, even if I maybe miss some advantages with LVM-groups and whatnot.
Anyway, I saw afterwards you were using a md0-device, ie raid. That may complicate things. 8-}
Jussi Hirvi wrote:
What should I do to make an existing CentOS (5.4) disc boot up on a new computer? [...] Would it be enough to boot with a DVD in rescue mode, or boot with another hd, and install grub?
On 16.12.2009 12:16, Sorin Srbu wrote:
For me it has worked to just install the old hd in the new machine and boot it up. Kudzo takes care of the rest.
Then you have been lucky. :-) For me, the startup stopped already before the CentOS splash screen. I guess something was wrong with the initrd.
If the disk holding the / partition needs a different driver than what you had during the install, you have to rebuild the initrd. Anaconda knows how to do that, kudzo can't. You can do it from a rescue-mode boot, but you may have to know the right module names.
At Wed, 16 Dec 2009 07:33:38 -0600 CentOS mailing list centos@centos.org wrote:
Jussi Hirvi wrote:
What should I do to make an existing CentOS (5.4) disc boot up on a new computer? [...] Would it be enough to boot with a DVD in rescue mode, or boot with another hd, and install grub?
On 16.12.2009 12:16, Sorin Srbu wrote:
For me it has worked to just install the old hd in the new machine and boot it up. Kudzo takes care of the rest.
Then you have been lucky. :-) For me, the startup stopped already before the CentOS splash screen. I guess something was wrong with the initrd.
If the disk holding the / partition needs a different driver than what you had during the install, you have to rebuild the initrd. Anaconda knows how to do that, kudzo can't. You can do it from a rescue-mode boot, but you may have to know the right module names.
*Before* swapping out the old disk, add an appropriate scsi_hostadapterN (N >= 1) alias to /etc/modprobe.conf and then do:
mkinitrd -f /boot/initrd-`uname -r`.img `uname -r`
All should be good then.
IF both the old machine and the new machine have your basic, vanila IDE disks, then there is no problem.
Robert Heller wrote:
At Wed, 16 Dec 2009 07:33:38 -0600 CentOS mailing list centos@centos.org wrote:
Jussi Hirvi wrote:
What should I do to make an existing CentOS (5.4) disc boot up on a new computer? [...] Would it be enough to boot with a DVD in rescue mode, or boot with another hd, and install grub?
On 16.12.2009 12:16, Sorin Srbu wrote:
For me it has worked to just install the old hd in the new machine and boot it up. Kudzo takes care of the rest.
Then you have been lucky. :-) For me, the startup stopped already before the CentOS splash screen. I guess something was wrong with the initrd.
If the disk holding the / partition needs a different driver than what you had during the install, you have to rebuild the initrd. Anaconda knows how to do that, kudzo can't. You can do it from a rescue-mode boot, but you may have to know the right module names.
*Before* swapping out the old disk, add an appropriate scsi_hostadapterN (N >= 1) alias to /etc/modprobe.conf and then do:
mkinitrd -f /boot/initrd-`uname -r`.img `uname -r`
All should be good then.
IF both the old machine and the new machine have your basic, vanila IDE disks, then there is no problem.
I've always wished the install/rescue disk had a mode to do this for you after you've moved the disks or restored a backup. The reason you are trying to bring up the new machine may be that the old one is dead - and anaconda knows a lot more about picking the right driver modules than I ever will. I've done it a time or two by installing a system on the new (or matching) hardware with a separate /boot partition, then making sure the old/new systems are updated to the same versions and keeping the new /boot but copying the rest of the old system over.
At Wed, 16 Dec 2009 09:55:49 -0600 CentOS mailing list centos@centos.org wrote:
Robert Heller wrote:
At Wed, 16 Dec 2009 07:33:38 -0600 CentOS mailing list centos@centos.org wrote:
Jussi Hirvi wrote:
What should I do to make an existing CentOS (5.4) disc boot up on a new computer? [...] Would it be enough to boot with a DVD in rescue mode, or boot with another hd, and install grub?
On 16.12.2009 12:16, Sorin Srbu wrote:
For me it has worked to just install the old hd in the new machine and boot it up. Kudzo takes care of the rest.
Then you have been lucky. :-) For me, the startup stopped already before the CentOS splash screen. I guess something was wrong with the initrd.
If the disk holding the / partition needs a different driver than what you had during the install, you have to rebuild the initrd. Anaconda knows how to do that, kudzo can't. You can do it from a rescue-mode boot, but you may have to know the right module names.
*Before* swapping out the old disk, add an appropriate scsi_hostadapterN (N >= 1) alias to /etc/modprobe.conf and then do:
mkinitrd -f /boot/initrd-`uname -r`.img `uname -r`
All should be good then.
IF both the old machine and the new machine have your basic, vanila IDE disks, then there is no problem.
I've always wished the install/rescue disk had a mode to do this for you after you've moved the disks or restored a backup. The reason you are trying to bring up the new machine may be that the old one is dead - and anaconda knows a lot more about picking the right driver modules than I ever will. I've done it a time or two by installing a system on the new (or matching) hardware with a separate /boot partition, then making sure the old/new systems are updated to the same versions and keeping the new /boot but copying the rest of the old system over.
Until very recently, I've moved disks from one AHA-29xxx SCSI system to another AHA-29xxx SCSI system. Same driver, different controller card... My latest move was on the same system, different disks: SCSI disks (AHA-29160N controller) to SATA (ahci flavored controller). In this case, the motherboard, etc. were working just fine (so where one of the *old* SCSI disks -- it died a couple of weeks later). The AHA-29160N controller card is still in the machine, with nothing connected to it (one never knows if some interesting piece of hardware happens along).
As for proper module: you just need to pay close attention to what anaconda is doing as it loads drivers. At least that is what I've seen. (I *always* use text mode with install/rescue disks.)
Thanks, Les, Robert et al. - good tips.
I did not have anaconda installed on these discs. I will do "yum install anaconda" at once. Maybe it will help the next time.
- Jussi
Oh, cancel the previous post... I will *not* install anaconda, I didn't remember it is an *installer* - I don't think it would help in this case.
This procedure by Les sounds realistic, though I wish there was an easier way:
I've done it a time or two by installing a system on the new (or matching) hardware with a separate /boot partition, then making sure the old/new systems are updated to the same versions and keeping the new /boot but copying the rest of the old system over.
On both my "old" (booting ok) and "new" (not booting) disc the modprobe.conf contained
alias scsi_hostadapter ata_piix
So it seems to me modprobe.conf was up-to-date.
- Jussi
Jussi Hirvi wrote:
Oh, cancel the previous post... I will *not* install anaconda, I didn't remember it is an *installer* - I don't think it would help in this case.
This procedure by Les sounds realistic, though I wish there was an easier way:
I've done it a time or two by installing a system on the new (or matching) hardware with a separate /boot partition, then making sure the old/new systems are updated to the same versions and keeping the new /boot but copying the rest of the old system over.
On both my "old" (booting ok) and "new" (not booting) disc the modprobe.conf contained
alias scsi_hostadapter ata_piix
So it seems to me modprobe.conf was up-to-date.
If the disks take the same driver it should boot or at least get to a point where you get an error message with the reason. Are you sure it isn't a cable or drive select problem?
Jussi Hirvi wrote:
On both my "old" (booting ok) and "new" (not booting) disc the modprobe.conf contained
alias scsi_hostadapter ata_piix
So it seems to me modprobe.conf was up-to-date.
On 17.12.2009 15:22, Les Mikesell wrote:
If the disks take the same driver it should boot or at least get to a point where you get an error message with the reason. Are you sure it isn't a cable or drive select problem?
I don't think there can be a cable select problem with SATA drives. :-)
Anyway, another drive booted ok when attached to the same SATA cable.
I did not get an error message, in fact the bootup hang on a blank screen (I have not seen that before) some time before the CentOS splash screen (I don't remember more exactly anymore, and cannot check).
- Jussi
Jussi Hirvi wrote:
Jussi Hirvi wrote:
On both my "old" (booting ok) and "new" (not booting) disc the modprobe.conf contained
alias scsi_hostadapter ata_piix
So it seems to me modprobe.conf was up-to-date.
On 17.12.2009 15:22, Les Mikesell wrote:
If the disks take the same driver it should boot or at least get to a point where you get an error message with the reason. Are you sure it isn't a cable or drive select problem?
I don't think there can be a cable select problem with SATA drives. :-)
Anyway, another drive booted ok when attached to the same SATA cable.
I did not get an error message, in fact the bootup hang on a blank screen (I have not seen that before) some time before the CentOS splash screen (I don't remember more exactly anymore, and cannot check).
The first boot stages should happen with the the machine bios loading grub, then the kernel and initrd. It sounds like you aren't even getting that far. The usual driver/controller mismatch results in the kernel still being loaded by bios, doing it's device detection, then failing because it can't mount the root partition. Is the disk you are trying to boot part of a raid? And maybe the wrong half?
At Thu, 17 Dec 2009 09:51:06 -0600 CentOS mailing list centos@centos.org wrote:
Jussi Hirvi wrote:
Jussi Hirvi wrote:
On both my "old" (booting ok) and "new" (not booting) disc the modprobe.conf contained
alias scsi_hostadapter ata_piix
So it seems to me modprobe.conf was up-to-date.
On 17.12.2009 15:22, Les Mikesell wrote:
If the disks take the same driver it should boot or at least get to a point where you get an error message with the reason. Are you sure it isn't a cable or drive select problem?
I don't think there can be a cable select problem with SATA drives. :-)
Anyway, another drive booted ok when attached to the same SATA cable.
I did not get an error message, in fact the bootup hang on a blank screen (I have not seen that before) some time before the CentOS splash screen (I don't remember more exactly anymore, and cannot check).
The first boot stages should happen with the the machine bios loading grub, then the kernel and initrd. It sounds like you aren't even getting that far. The usual driver/controller mismatch results in the kernel still being loaded by bios, doing it's device detection, then failing because it can't mount the root partition. Is the disk you are trying to boot part of a raid? And maybe the wrong half?
Or simply not marked as a bootable disk?
-
Jussi Hirvi -
Jussi Hirvi -
Les Mikesell -
Robert Heller -
Sorin Srbu