Linux Archive

Linux Archive (http://www.linux-archive.org/)
-   Fedora Development (http://www.linux-archive.org/fedora-development/)
-   -   anacanda: should we ignore the bios raid information on a disk when the raid is broken? (http://www.linux-archive.org/fedora-development/712377-anacanda-should-we-ignore-bios-raid-information-disk-when-raid-broken.html)

Hans de Goede 10-15-2012 08:06 AM

anacanda: should we ignore the bios raid information on a disk when the raid is broken?
 
Hi,

On 10/15/2012 09:23 AM, Joshua C. wrote:

I have a broken fake raid on my machine (intel p67 chipset with one of
the disks missing) and when trying to install F17 yesterday (with
up-to-date respin done with pungi) I was greeted with the following
message "disk sdXXX has bios raid information and..... blah..... is
part of a broken raid, ignoring sdXXX". After ignoring the message
later on I wasn't given the chance to use the spare disk.

I thought of patching anaconda to ignore the bios-raid-information and
to allow me to use the disk as I single HDD but I was wondering if
there are any side effects out of this?


Yes, the side effect of this is that if we wrongly detect an array as being
broken and allow the user to use it, we will destroy the array, nuking any
data on it. IOW ignoring this error is simply not an acceptable option.

What you can do is remove the bios raid metadata from the disk by going
into a rescue shell on the system and run wipefs on the disk in question

Regards,

Hans
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel

"Joshua C." 10-15-2012 08:41 AM

anacanda: should we ignore the bios raid information on a disk when the raid is broken?
 
2012/10/15 Hans de Goede <hdegoede@redhat.com>:
> Hi,
>
>
> On 10/15/2012 09:23 AM, Joshua C. wrote:
>>
>> I have a broken fake raid on my machine (intel p67 chipset with one of
>> the disks missing) and when trying to install F17 yesterday (with
>> up-to-date respin done with pungi) I was greeted with the following
>> message "disk sdXXX has bios raid information and..... blah..... is
>> part of a broken raid, ignoring sdXXX". After ignoring the message
>> later on I wasn't given the chance to use the spare disk.
>>
>> I thought of patching anaconda to ignore the bios-raid-information and
>> to allow me to use the disk as I single HDD but I was wondering if
>> there are any side effects out of this?
>
>
> Yes, the side effect of this is that if we wrongly detect an array as being
> broken and allow the user to use it, we will destroy the array, nuking any
> data on it. IOW ignoring this error is simply not an acceptable option.
>
> What you can do is remove the bios raid metadata from the disk by going
> into a rescue shell on the system and run wipefs on the disk in question
>
> Regards,
>
> Hans

I don't want to remove the bios data because this is the only way to
rebild the raid when the next disk arrives. Currently I'm using the
disk under Linux/Windows without any problems (in AHCI mode). Wipping
the bios data will remove anything when later I build (re-build) the
raid with the intel orom...

Can I just install anaything on the second disk and then manually
adjust the fstab file to automount the disk from the broken raid? The
raid is for my home partition.

--joshua
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel

Hans de Goede 10-15-2012 05:06 PM

anacanda: should we ignore the bios raid information on a disk when the raid is broken?
 
Hi,

On 10/15/2012 10:41 AM, Joshua C. wrote:

2012/10/15 Hans de Goede <hdegoede@redhat.com>:

Hi,


On 10/15/2012 09:23 AM, Joshua C. wrote:


I have a broken fake raid on my machine (intel p67 chipset with one of
the disks missing) and when trying to install F17 yesterday (with
up-to-date respin done with pungi) I was greeted with the following
message "disk sdXXX has bios raid information and..... blah..... is
part of a broken raid, ignoring sdXXX". After ignoring the message
later on I wasn't given the chance to use the spare disk.

I thought of patching anaconda to ignore the bios-raid-information and
to allow me to use the disk as I single HDD but I was wondering if
there are any side effects out of this?



Yes, the side effect of this is that if we wrongly detect an array as being
broken and allow the user to use it, we will destroy the array, nuking any
data on it. IOW ignoring this error is simply not an acceptable option.

What you can do is remove the bios raid metadata from the disk by going
into a rescue shell on the system and run wipefs on the disk in question

Regards,

Hans


I don't want to remove the bios data because this is the only way to
rebild the raid when the next disk arrives. Currently I'm using the
disk under Linux/Windows without any problems (in AHCI mode). Wipping
the bios data will remove anything when later I build (re-build) the
raid with the intel orom...

Can I just install anaything on the second disk and then manually
adjust the fstab file to automount the disk from the broken raid?


Assuming your raid array is a mirror, and that you won't be partitioning
it or something similar, just adding the existing /home partition to
your fstab yes that should work. Although anaconda will not let you
touch the mirror member during the install, if you've another disk,
putting Fedora 17 on that other disk should work fine, and after
that pointing fstab the disk will work.

BUT *IMPORTANT IMPORTANT IMPORTANT* you MUST remove the entry from
fstab, before rebuilding the array, and then after the rebuild re-add
the entry put now pointing to the raid and not to the single disk,
otherwise Linux will keep using the single disk for your /home !!!

Regards,

Hans
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel


All times are GMT. The time now is 12:22 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.