Description of problem:
If you try to install (not upgrade) over the top of existing raid
devices, the anaconda install will fail with a python exception. This
can be worked around if in the %pre scriptlet one zeros the raid superblock.
It would be desirable to have a feature whereby one could tell anaconda to
just ignore the existing raid devices. Something like:
would be great.
The problem was discussed in the issue tracker at:
And in the following emails to the kickstart list:
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Install a system with raid using kickstart.
2. Run the same kickstart on the same system.
You will receive a python exception.
The install should have reinstalled just fine.
By George, I think I've got it. In partitions.py, there are
if tmp and tmp.type == REQUEST_RAID:
boot = 
for member in tmp.raidmembers:
boot = [tmp]
boot = 
My attempt to debug the process show that the
line makes boot be an array of None types
(with as many Nones as partitions in the raid) and
that it is because the IDs stored in tmp.raidmembers
don't correspond to any actual partition IDs in self.
Why that is happening, I don't know yet.
Is there a way in KickStart to make Anaconda launch the
debugger at the very beginning? It would really come in handy.
I'm new to this bug and it's been a while without any activity, so I'll ask the
obvious question. Have you tried this against a recent Rawhide, Fedora Core, or
RHEL? There's been a pile of changes so it's hard to say whether or not this
has been inadvertently fixed since the bug was originally reported.
*** Bug 149992 has been marked as a duplicate of this bug. ***
Going to try it this weekend.(In reply to comment #2)
> I'm new to this bug and it's been a while without any activity, so I'll ask the
> obvious question. Have you tried this against a recent Rawhide, Fedora Core, or
> RHEL? There's been a pile of changes so it's hard to say whether or not this
> has been inadvertently fixed since the bug was originally reported.
Going to try with Fedora Core 3 this weekend.
Well it still crashes with FC3, see the bug which has just been closed as a
duplicate of this one.
I have a potential fix for this in testing here. I believe it's caused by the
same problem as 156283. Stay tuned.
Committed a potential fix to CVS. Please try with fc4 when available and let me
This problem has been fixed in Rawhide.