Bug 225551
Summary: | bootrecord always written to MBR, when dmraid is installed | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Winfrid Tschiedel <Winfrid.Tschiedel> | ||||||||||
Component: | anaconda | Assignee: | Joel Andres Granados <jgranado> | ||||||||||
Status: | CLOSED RAWHIDE | QA Contact: | |||||||||||
Severity: | medium | Docs Contact: | |||||||||||
Priority: | medium | ||||||||||||
Version: | 8 | ||||||||||||
Target Milestone: | --- | ||||||||||||
Target Release: | --- | ||||||||||||
Hardware: | x86_64 | ||||||||||||
OS: | Linux | ||||||||||||
Whiteboard: | |||||||||||||
Fixed In Version: | Doc Type: | Bug Fix | |||||||||||
Doc Text: | Story Points: | --- | |||||||||||
Clone Of: | Environment: | ||||||||||||
Last Closed: | 2008-04-07 15:41:22 UTC | Type: | --- | ||||||||||
Regression: | --- | Mount Type: | --- | ||||||||||
Documentation: | --- | CRM: | |||||||||||
Verified Versions: | Category: | --- | |||||||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||||||
Embargoed: | |||||||||||||
Bug Depends On: | |||||||||||||
Bug Blocks: | 236452 | ||||||||||||
Attachments: |
|
Description
Winfrid Tschiedel
2007-01-31 10:09:36 UTC
Looks more like anaconda issue to me than dmraid. This bug still exists on Fedora 8 - Fedora 9 Alpha I could not test, because I could install the first release of Fedora 9, I get a backtrace in the installer but I am not able to save this information. This error exists also on rhel 5.1 ! There is a ks file produces by anaconda that you can retrieve from your system. It is located in /root/anaconda-ks.cfg. the files for both systems would help. thx. I really dont think that this is relsated to raid. Here are my tests and the results: Setup: Have two scsi drives, x86_64 arch 1. installed f8 in one drive, chose the erase everything and installed anacondas default setup in one of the drives. Created a boot partition that is used for booting. 2. installed another f8 system in the other drive, but this time I took care of setting up the second system to boot from the /boot partition on the other disk. Both installations succeeded. The grub.conf cont overwriten by the grub backage (I assume). The information of the previous probrama is in the grub.conf.rpmsave file. Are you seeing the same? Created attachment 294914 [details]
requested anaconda-ks.cfg from fedora 8
if file for rhel 5.1 is also needed I have make a new installation,
because the system where I observed the problem is not anymore available.
Winfrid
(In reply to comment #5) > Created an attachment (id=294914) [edit] > requested anaconda-ks.cfg from fedora 8 > > if file for rhel 5.1 is also needed I have make a new installation, ^^^^^^^^ this would be very helpfull. :) Q1: do you have two disks or are you installing in two different partitions? Q2: Are you doing a kickstart install or normal graphical one? Q3: In your original test did you use the /boot partition to boot? (In reply to comment #4) > I really dont think that this is relsated to raid. Here are my tests and the > results: > Setup: Have two scsi drives, x86_64 arch > 1. installed f8 in one drive, chose the erase everything and installed anacondas > default setup in one of the drives. Created a boot partition that is used for > booting. > 2. installed another f8 system in the other drive, but this time I took care of > setting up the second system to boot from the /boot partition on the other disk. > Both installations succeeded. The grub.conf cont overwriten by the grub > backage (I assume). The information of the previous probrama is in the > grub.conf.rpmsave file. > Are you seeing the same? My operation is a little bit different - if I have a new system, then I start this way. Install first OS to /boot + / After the first installation has finished, I copy the boot-partition to the root-partition and do all changes in the new boot-partition in order to be able to boot this system from the root partition. This means I have to create a boot record for this root partition, after this I adjust /etc/fstab and the new "/boot/grub/menu.lst". After this I change grub/menu.lst on the old boot partition and do chain loading of all installed linux distributions. After this install all other distributions in seperate partition and write the boot-record to the partition. As long as I was using non raid devices only in some rare cases the mbr was overwritten ( no bootrecord in the partition )- this happens mostly for beta releases. In case of raid devices ( dmraid ) always a mbr is created ( in case of rhel/fedora without warning ), in case of openSUSE, you get a warning, which tells you, that the br is written to default location. (In reply to comment #6) > (In reply to comment #5) > > Created an attachment (id=294914) [edit] [edit] > > requested anaconda-ks.cfg from fedora 8 > > > > if file for rhel 5.1 is also needed I have make a new installation, > ^^^^^^^^ > this would be very helpfull. :) > Q1: do you have two disks or are you installing in two different partitions? > Q2: Are you doing a kickstart install or normal graphical one? > Q3: In your original test did you use the /boot partition to boot? I think I answered some of your questions in the meantime, but here is the direct answer to your questions: Q1 : different partitions - but see my remark Q2 : graphical installation Q3 : only for the first system I have a seperate boot partition, but this is just used for the installation, all other systems are installed into one partition. Remark : Currently I have a second problem with dmraid and fedora 8 - My system has a Promise PDC20319 Controller (FastTrak S150 TX4) (rev 02). In this system I have to define every disk as a RAID device, otherwise I am not able to boot from this device. So I have in my system 2 RAID0 ( with just 1 disk ). In case of fedora and 2 disks anaconda (?) loops, so I never get to the disk partitioning. (In reply to comment #7) > My operation is a little bit different - .... > I adjust /etc/fstab and the new "/boot/grub/menu.lst". I see, did this work in previous versions of fedora/rhel? Ok, I tried to mirror what you are trying to do and came up with a solution. I'm assuming that you are using the extended bootloader screan and telling anaconda to put the bootloader in the current disk (refering to the second install, the one you don't want to have a bootloader). If you do this your actually telling anaconda to put the bootloader there. What you should do to get the behavior you want is to select the "not install bootloader" on the second install. In this way you install a system that is not bootable. To make the system bootable you take the initrd.img and vmlinuz from /boot and put it in the /boot directory from the first system you installed. You also must change the grub.conf to reflect the changes. If you get everything right the second system will be included in the grub list and will be bootable :) Created attachment 295008 [details]
requested anaconda-ks.cfg from rhel 5.1
+ output from raid -r :
/dev/sda: pdc, "pdc_cjgdhdfafh", stripe, ok, 156118928 sectors, data@ 0
Created attachment 295019 [details]
anaconda-ks fromfedora 8 ( fresh install )
Today I installed my fedora 8 into another partition ( on disk 1 ) -
previous installation was on disk 2 ( install on 2 disk, disk 1 offline ).
requested output from dmraid -r :
/dev/sda: pdc, "pdc_bjfeeibeeb", stripe, ok, 312368928 sectors, data@ 0
/dev/sdb: pdc, "pdc_cjgdhdfafh", stripe, ok, 156118928 sectors, data@ 0
Created attachment 295020 [details]
anaconda-ks fromfedora 8 ( fresh install )
Today I installed my fedora 8 into another partition ( on disk 1 ) -
previous installation was on disk 2 ( install on 2 disk, disk 1 offline ).
requested output from dmraid -r :
/dev/sda: pdc, "pdc_bjfeeibeeb", stripe, ok, 312368928 sectors, data@ 0
/dev/sdb: pdc, "pdc_cjgdhdfafh", stripe, ok, 156118928 sectors, data@ 0
Please have also a look at issue 164550 on enterprise.redhat.com/issue-tracker IT#164550 and the original problem described in this Bug is not the same. Please treat as two separate issues. /Anders This was solved in booty. bug number (#236452) |