[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#79890: Package boot-floppies fails to install om Compaq Smart Raid



"Magnus Pettersson" <magnus@easyliving.com> writes:

> Package: boot-floppies
> Version: 2.2.17 (potato2.2-2)
> 
> I tried to install Potato 2.2R2 on a Compaq 6400 server with 2 Compaw Smart
> Raid controllers, one integrated (221) and one externel (4200).
> 
> Problem:
> History:On potato2.2R0 the installation broke when adding a second
> controller, and the system was only able to install on one of the
> controllers. (this of course using the 2nd CD with the 'compact' kernel.
> since CD1 doesn't recognize the Compaq SMART arraycontroller (as it
> should)).
> However on Potato 2.2R2 it breaks completely, Finding all SMART controllers,
> but failing to create ANY dev files for them (/dev/ida/cXd0...), and of
> course therefor not being able to install on any drive.
> 
> Solution:
> I believe the error lies in the number of /dev-files the system is able to
> create during installation (due to limits of the ramdisk?), since I was able
> to custom-build a potato2.2R2 CD from my local mirror editing rescue.sh
> (boot-floppies) and change the code, so that  MAKEDEV did not create all of
> the /dev-files it normally does (I made it only create the /dev-files I
> needed to install). When I booted that custom-made CD, all /dev-files for
> both of the SMART controllers was created, and I was able to install on both
> SMART arraycontrollers.

Exactly which files in /dev are not there that should be there?  

On i386, I'm pretty sure, adding some more files in /dev is not a big
problem, but we do try to keep it under control.

> Note:
> When making the changes to rescue.sh,

Um, wouldn't it be a patch to rootdisk.sh ?

> fixes that CRITICAL bug, there is a
> minor bug also, when under the installation, you are asked to 'make linux
> bootable direct from hard-disk':

> Problem2:

FYI, it's better to keep it to one problem per bug report.

> I had installed my system with root on /dev/ida/c1d0p1 and wanted to install
> an MBR on /dev/ida/c1d0 and the installation only allowed me to install the
> MBR on /dev/ida/c0d0 OR /dev/ida/c1d0p1, both of course wrong.

Ouch.  This is another bug in libfdisk I think.  I'm not really
skilled to fix this.  Can anyone try to fix it and send a patch to
this bug report ?


-- 
.....Adam Di Carlo....adam@onShore.com.....<URL:http://www.onShore.com/>




Reply to: