Mailinglist Archive: opensuse-features (82 mails)

< Previous Next >
[openFATE 305099] please have yast do a fsck before attempting to mount existing filesystems
Feature changed by: Christoph Thiel (cthiel1)
Feature #305099, revision 12
Title: please have yast do a fsck before attempting to mount existing

- openSUSE-11.2: Rejected by Marcus Kraft (cthiel1)
+ openSUSE-11.2: Rejected by Christoph Thiel (cthiel1)
reject date: 2009-10-27 16:03:38
reject reason: postponed.
Requester: Important
Projectmanager: Desirable

openSUSE-11.3: Evaluation by product manager
Requester: Important

Requested by: Per Jessen (pjessen)
Product Manager: Federico Lucifredi (flucifredi)
Partner organization:

I was doing a new install, but wanted to keep my /home filesystem
(JFS). No problem, except that the filesystem wasn't clean and needed
an fsck. As far as I can tell, yast didn't do this, so mounting of the
/home filesystem failed, which caused the entirely installation process
to abend and start over.

- please have yast do a fsck before attempting to mount existing
filesystems (novell/bugzilla/id: 361728)

#1: Stefan Hundhammer (sh-sh-sh) (2008-02-21 06:26:43)
I'd really hate to have my 100+ GB ext3 partitions fscked. That takes

#2: Per Jessen (pjessen) (2008-02-21 07:15:33)
Even when they're clean? It's done on every bootup anyway. See

#3: Jozef Uhliarik (juhliarik) (2008-02-21 08:21:41)
What is your opinion Thomas?

#4: Thomas Fehr (fehr) (2008-02-21 08:24:13)
That would need an feature request, not a bug report.

#6: Stephan Kulow (coolo) (2008-02-21 12:04:27)
I think it's not too uncommon that people reinstall their system
_because_ the old one had a problem, so dirty file systems may not be
too uncommon.

#7: Thomas Fehr (fehr) (2008-02-25 03:00:03)
What the customer described happens on JFS, this is an unsupported
filesystem exactly for the reason that it behaves quite fundamentally
different to most other filesystems.
Filesystems like ext3 or reiser simply do their log reply when they are
dirty and all is fine. So if you want such things to be added, please
specify it as a feature request. Since there needs to be an interactive
user interface to the filesystem check program this is in no way a
simple change. Or would you prefer for example do a rebuild of reiserfs
internal tree as default without asking the user first? Problem is that
the questions the filesystem check tools ask the user are quite special
and of course totally different for different filesystem types.

#10: Robert Davies (robopensuse) (2009-01-16 16:51:27) (reply to #7)
JFS fsck.jfs -p seems to make sense. ext2 fsck.ext2 suports that to.
Validating filesystems that are not to be formatted, is a reasonable
thing to do (it happens on boot after all) so is not against user
expectations. You just booted to do the installation after all. As for
"non trivial" UI, why is aborting earlier only when recovery is
impossible, worse than failing with zero effort on the mount?

#8: Duncan Mac-Vicar (dmacvicar) (2008-07-17 08:13:00)
I happen to suffer from this bug too once. I think the bug is not that
the installation cleans the disk, but at least say that is not clean
instead of crashing (in my case it was an installation where yast
thought the disk was mounted but it was not)

#9: Arvin Schnell (aschnell) (2008-07-17 08:22:23)
No time for 11.1.

#11: Jan Engelhardt (jengelh) (2009-06-18 23:55:19)
Not mounting is one thing, but cancelling the installation is

#12: Robert Davies (robopensuse) (2009-11-12 11:33:32)
You get aborted installation similarly if an invalid mount parameter is
Can't trial existing mounts be tried after partition info is entered
(with fsck -p possibility if the mount fails in such a way to make fsck
recommended).  Then the installer can be guided back to adjust
partition info & options, rather than have it fail much later when
beginning the installation.
A check box for "Perform Consistency Check" could be set, with
suggestion to use the "Do not mount at boot" option to repair it later,
if fsck-ing is unwanted.

openSUSE Feature:

< Previous Next >
This Thread
  • No further messages