Bugzilla Bug 51773

mkfs.reiserfs causes major problems.

I created a filesystem on my new SATA using mkfs.reiserfs copied a large tarball to it, and untarred it.  I got major errors.  The file system was then corrupted and all data was lost.

I repeated the exact same procedure using mkreiserfs instead this time, and I had no problem.

I repeated again with mkfs.reiserfs and then again with mkreiserfs to make sure it was repeatable.  It is repeatable and I have come to the conclusion that mkfs.reiserfs is critically unstable.  I think it should be removed from the sys-fs/progsreiserfs ebuild, or the entire progsreiserfs ebuild should be deprecated if it is out of date or something.

Just want to note, that qtparted uses progsreiserfs, too and I lost data
because of it, trying to resize a partition, so I concur David.

Brandon are you there?

Can this be resolved by a developer?  Can something be done immediately?

Have a look at http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=244169, too.

Hell ya, that Debian bug is bang on.

Please Gentoo people let's take the initiative and completely remove this buggy piece of software from portage completely.  Or make it ~x86 and put a huge warning in the pkg_setup or whatever.  Make it beep three times too...

ok, i package masked it and removed it from the DEPEND of parted/qtparted

there isnt anything i can do about it already being installed on your system though :/

Should't there be an announcement for base-system packages that cause so fatal
errors?

it's not a 'base system package'

it's a package that either (1) was emerged directly by the user or (2) was pulled in because of parted

base-system herd just manages it because it's filesystem related

Oh, I thought it is. But it still can trash others fs. I mean, the package was
declared stable and even people who trust the Gentoo stable tree, don't even
get information about the fact that the package is masked know for security
reasons. Maybe Portage should be enhanced to trigger a warning in such cases.
But up to now it is the developers job, to make sure everyone will get the
information.

Next problem: qtparted needs --disable-reiserfs otherwise configure fails.
-> http://forums.gentoo.org/viewtopic.php?t=187231

Argh - now, not know, of course.

yep, fixed, sorry

does this happen with the 0.3.1rc8 version of progsreiserfs as well as the last
version in portage?

Didn't tested the snapshots, Robin. But the last one (0.3.1rc8) is from
12-Dec-2002 - seems to be a dead project.

a warning to all here.
progsreiserfs has been brought back, so that we can try and use the functionality that does work in it.
testdisk is the main application for this (it's able to recover data from damaged reiserfs filesystems). For the time being, it would be best if *parted did not use progsreiserfs at all. parted seems to be fine on this, but I can't test out qtparted properly here.

and to add to that, the new progsreiserfs doesnt install the binaries anymore,
just the library

SpanKY,
The new ebuild has:

make install DESTDIR="${D}" || die "Install failed"
rm -r "${D}"/usr/{sbin/share/man}

I don't see any || die on the rm, and I'm wondering if we should have such a thing there. Is there are safer way to do this?

no, the ebuild says:
rm -r "${D}"/usr/{sbin,share/man}

yeah, putting a || die on there would be nice, i'll add that to cvs