[Lunar-bugs] [Lunar Linux 0000067]: unable to create software-RAID with 1.6.0

Lunar bug reports list lunar-bugs at lunar-linux.org
Thu Feb 23 18:31:32 UTC 2006


A NOTE has been added to this issue.
======================================================================
<http://bugs.lunar-linux.org/view.php?id=67> 
======================================================================
Reported By:                remsys
Assigned To:                sofar
======================================================================
Project:                    Lunar Linux
Issue ID:                   67
Category:                   ISO
Reproducibility:            always
Severity:                   minor
Priority:                   normal
Status:                     assigned
Moonbase Version:           20060131
Core Tools:                 Lunar
Core Tools Version:         20060129
======================================================================
Date Submitted:             02-22-2006 11:35 UTC
Last Modified:              02-23-2006 18:31 UTC
======================================================================
Summary:                    unable to create software-RAID with 1.6.0
Description: 
I can't get my machines to setup a software-RAID, not through the
installer, nor from the command line in a temporary shell.

Here's what goes wrong:

INSTALLER
=========
Partitioning:
-------------
The first thing that strikes me as odd is that I can's see /dev/hdc if I
try to partition my disks, although it's definitely there. So I have to
exit to a tempshell to partition /dev/hdc.

In regard to a software-RAID it would be nice/useful to have the
possibility to copy the partition-scheme of one drive to another.

Setup RAID:
-----------
First of all I'd like the installed to warn me if I'm not using FS-type fd
on the disks I'm trying to turn into an array.

Secondly the installer is unable to create and start a raid-device (i.e.
/dev/md0) if it already contains a superblock. it should use
--really-force or at least ask you if you'd like to do so.

If I try to start any other array than /dev/md0 (i.e. /dev/md1 having
members /dev/hda2 and /dev/hdc2, which have both been setup correctly) the
installer simply restarts all over again!

trying to setup&start all other arrays from the commandline in a
tempshell, as I have done many times before with the 2.4-based kernels,
doesn't work either, simply because there aren't any other /dev/md devices
than /dev/md0 -> /dev/md/0 (???????????). I haven't got a clue how to get
those, so I'll roll back to the 1.5.2 iso......
======================================================================

----------------------------------------------------------------------
 sofar - 02-23-06 15:31 
----------------------------------------------------------------------
> The first thing that strikes me as odd is that I can's see /dev/hdc if I
try to
> partition my disks, although it's definitely there. So I have to exit to
a
> tempshell to partition /dev/hdc.

this was fixed in SVN already as reported on the ML.

----------------------------------------------------------------------
 sofar - 02-23-06 15:32 
----------------------------------------------------------------------
> First of all I'd like the installed to warn me if I'm not using FS-type
fd on
> the disks I'm trying to turn into an array.

I'll consider this but it's not trivial to implement

----------------------------------------------------------------------
 sofar - 02-23-06 15:34 
----------------------------------------------------------------------
> If I try to start any other array than /dev/md0 (i.e. /dev/md1 having
members
> /dev/hda2 and /dev/hdc2, which have both been setup correctly) the
installer
> simply restarts all over again!

> trying to setup&start all other arrays from the commandline in a
tempshell, as
> I have done many times before with the 2.4-based kernels, doesn't work
either,
> simply because there aren't any other /dev/md devices than /dev/md0 ->
> /dev/md/0 (???????????). I haven't got a clue how to get those, so I'll
roll
> back to the 1.5.2 iso......

I can't reproduce this - the installer currently creates the /dev/md*
device nodes manually - I'll re-visit the code for testing for the next
release anyway, so this will get tested (again)

----------------------------------------------------------------------
 sofar - 02-23-06 15:36 
----------------------------------------------------------------------
> the installer is unable to create and start a raid-device (i.e. /dev/md0)
if it
> already contains a superblock. it should use --really-force or at least
ask you
> if you'd like to do so.

I don't agree with this - I'm not going to help people destroy their data
;^)

----------------------------------------------------------------------
 remsys - 02-23-06 18:27 
----------------------------------------------------------------------
>> If I try to start any other array than /dev/md0 (i.e. /dev/md1 having
members
>> /dev/hda2 and /dev/hdc2, which have both been setup correctly) the
installer
>> simply restarts all over again!

>> trying to setup&start all other arrays from the commandline in a
tempshell, as
>> I have done many times before with the 2.4-based kernels, doesn't work
either,
>> simply because there aren't any other /dev/md devices than /dev/md0 ->
>> /dev/md/0 (???????????). I haven't got a clue how to get those, so I'll
roll
>> back to the 1.5.2 iso......

> I can't reproduce this - the installer currently creates the /dev/md*
device 
> nodes manually - I'll re-visit the code for testing for the next release

> anyway, so this will get tested (again)

Hmmm, it might have something to do with the fact that md1 (hda1+hdc1)
already contained a superblock as well(?). I wasn't able to figure that
out because of timepressure.

----------------------------------------------------------------------
 remsys - 02-23-06 18:31 
----------------------------------------------------------------------
>> the installer is unable to create and start a raid-device (i.e.
/dev/md0) if it
>> already contains a superblock. it should use --really-force or at least
ask you
>> if you'd like to do so.

> I don't agree with this - I'm not going to help people destroy their
data ;^)

hmmm, but it would be nice to be able to actually start the array from the
installer if you know what you are doing ("would you like me to force
recreation of this array and DESTROY THE DATA IN IT?"), especially for
people that don't know their way around in the installer.....

Issue History
Date Modified  Username       Field                    Change              
======================================================================
02-22-06 11:35 remsys         New Issue                                    
02-22-06 11:35 remsys         Moonbase Version          => 20060131        
02-22-06 11:35 remsys         Core Tools                => Lunar           
02-22-06 11:35 remsys         Core Tools Version        => 20060129        
02-23-06 15:31 sofar          Note Added: 0000118                          
02-23-06 15:31 sofar          Severity                 block => minor      
02-23-06 15:32 sofar          Note Added: 0000119                          
02-23-06 15:34 sofar          Note Added: 0000120                          
02-23-06 15:36 sofar          Note Added: 0000121                          
02-23-06 18:27 remsys         Note Added: 0000122                          
02-23-06 18:31 remsys         Note Added: 0000123                          
======================================================================



More information about the Lunar-bugs mailing list