Re: Bug#651624: Booting from zfs root seems to not work 8.3 and 10.0 however work
- To: Christoph Egger <christoph@debian.org>
- Cc: 651624@bugs.debian.org, debian-bsd@lists.debian.org
- Subject: Re: Bug#651624: Booting from zfs root seems to not work 8.3 and 10.0 however work
- From: Robert Millan <rmh@debian.org>
- Date: Sun, 25 Mar 2012 23:26:14 +0200
- Message-id: <[🔎] CAOfDtXNo2oVTG4i+1yavs0NncTSpU3eHL+gwVCCwFC3LmYre1g@mail.gmail.com>
- In-reply-to: <[🔎] 87limoe6ew.fsf@hepworth.siccegge.de>
- References: <20111210165258.2773.95763.reportbug@mitoraj.siccegge.de> <CAOfDtXO=J62J0vUoHJ7+J6UwuEymCX5opYeo9VLydoUQQuE21Q@mail.gmail.com> <87zkf0491g.fsf@hepworth.siccegge.de> <87borg2oyp.fsf@hepworth.siccegge.de> <CAOfDtXPATg=QH1C+tfPrgD1JRXSBsBTTAbs5yr-kq8zR2FuGFQ@mail.gmail.com> <20111212122759.GA30687@oteiza.siccegge.de> <CAOfDtXOM3rChO+qE0gq1gWuGkVJi6d6ms_0LsLM=VSW+5rv4gw@mail.gmail.com> <87sjkpz9oa.fsf@hepworth.siccegge.de> <CAOfDtXP1i55kieCn9JHD=yS6CJOz-ivfh7_OhOWAaBNbEPX-CA@mail.gmail.com> <20111221112301.GA15101@oteiza.siccegge.de> <CAOfDtXMPotYdjUDKEMFFGo2GVOtGdwup1AnJaKySZV2Z4vshew@mail.gmail.com> <8739alm8zi.fsf@hepworth.siccegge.de> <CAOfDtXOwr_oHCgdbOv2093t0Wg7yedAVeK-FqW_JqouQ22_boQ@mail.gmail.com> <[🔎] 87limoe6ew.fsf@hepworth.siccegge.de>
El 25 de març de 2012 23:07, Christoph Egger <christoph@debian.org> ha escrit:
> Robert Millan <rmh@debian.org> writes:
>> I'm not sure if people in debian-bsd can help with this. I myself
>> can't. Maybe you should try to reproduce this in a pure FreeBSD
>> environment by trying to import the pool from a FreeBSD system. If it
>> can be imported with kFreeBSD 8.x (either GNU or BSD userland), but
>> not with pure FreeBSD 9.0, then it's IMHO worth reporting to ZFS
>> experts in freebsd-fs.
>
> I was now able to reproduce the problem in a 10-CURRENT build done on
> pure freebsd 9.0-STABLE with default cc though the failing box is (of
> course) still kfreebsd.
I was hit by this too. Really annoying :-(
> There should be no othyer debian influcenes before the reoot filesystem
> comes available apart from grub right? Going to take this upstream if
> so.
There might be something written on disk by Debian kernel that
upstream kernel doesn't. Probably not file system corruption since
it's clearly recoverable (as I write, my system just came back to
normal after simply booting a D-I image and importing the pool there).
But as the data is there, if upstream kernels can't access it I think
it's probably relevant to them.
If you bring this upstream, I recommend writing to freebsd-fs, it's
much easier to get attention there than by filing a PR.
--
Robert Millan
Reply to: