Re: boot failure
- To: debian-user@lists.debian.org
- Subject: Re: boot failure
- From: Marc Auslander <marcausl@gmail.com>
- Date: Wed, 25 Jan 2012 10:13:21 -0500
- Message-id: <[🔎] 87d3a7erxq.fsf@aptiva.optonline.net>
- In-reply-to: <inaiS-7NR-15@gated-at.bofh.it> (Pierre Frenkiel's message of "Wed\, 25 Jan 2012 13\:40\:02 +0100")
- References: <imNmh-2D3-5@gated-at.bofh.it> <imNPk-3h5-13@gated-at.bofh.it> <imPeq-5Oz-15@gated-at.bofh.it> <imQav-7qI-33@gated-at.bofh.it> <imSvF-2Rw-23@gated-at.bofh.it> <inaiS-7NR-15@gated-at.bofh.it>
Pierre Frenkiel <pierre.frenkiel@gmail.com> writes:
> On Tue, 24 Jan 2012, Pierre Frenkiel wrote:
>
> Remains the question "how /etc/modprobe.d/options was corrupted"
> May-be after a power failure, but I thought that this couldn't
> happen with a ext4 file system.
>
The default/normal configuration of ext4 journals only metadata.
This means that file system integrity is guaranteed but individual
file content can be corrupted by a power failure or other hard crash.
To update files that must be correct, one can write a new version,
sync, and then use mv to replace the old with the new version. Since
mv is a metadata operation, it is guaranteed to either have happened
or not after crash recovery.
The journal_data option provides full protection, but I suspect it is
not a practical solution for normal usage.
Reply to: