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

partman problems (sparc and elsewhere)

After doing successful installs using rc2 on a stack of sparc systems
(ultra1, ultra1e, ultra2, ultra2) I started playing with raid and lvm
installs.  Unfortunatly, the raid and lvm stuff failed due to bugs in
partman and missing modules.

However, I did find a counterintuitive workaround for bug 239300:
(which I duplicated on an ultra2) choose the raid install option.  It
fails, but makes the free partition visable.

Looking into the partman bugs reports, I think all of these bugs are
symptoms of the same partman bug:

	239441	(merged)
	246980	(merged)
	258880	(merged)
	263208	(merged)
	268580	(merged)
and maybe

Bug 279017 describes what partman is messing up best.

My summery is that partman has information in memory on what the
desired partitioning is, and information on disk.  It can write
information to the disk without updating all of its in memory
information to match, and gets itself into an inconsistant state.
Choosing another menu item that fails totaly can get it back to a
consistant state.  Choosing a menu item that partly succeds (such as
lvm) can make things even crazier.

This isn't a hard to reproduce bug, pretty much the only case that
works is a simple single-disk install with ext3 partitons.

Blars Blarson			blarson@blars.org
With Microsoft, failure is not an option.  It is a standard feature.

Reply to: