"Stephen C. Tweedie" wrote:
> Hi,
>
> On Wed, Apr 05, 2000 at 03:11:44PM -0400, Lyle Seaman wrote:
> >
> > How does a readonly filesystem become inconsistent?
> > (esp: "how does ext3 on readonly media" become inconsistent?)
> >
> > The obvious answer is "well, it *wasn't* readonly when it
> > became inconsistent".
> >
> > If that's the case, then why do you care? Naively, I wouldn't
> > think this is a big deal. Why am I wrong?
>
> Because (a) existing Linux installations expect to mount the
> root filesystem readonly until basic consistency checking has
> been done; and (b) after a cold reboot, the filesystem will
> need recovery before it can be mounted.
>
> In practice there is no problem mounting the root filesystem
> read-write if you have journaling, with one important exception:
> if you detect actual errors on the fs such that a constency
> check is required, mounting the fs read-write is dangerous.
> (Ext2/ext3 mark an error flag in the superblock if they detect
> dangerous inconsistencies in the fs so that a subsequent fsck
> will be forced. I don't know if xfs does this or not.)
>
XFS doesn't actually try to detect file system inconstancies at mount
time.
(this would defeat quick recovery)
If it can successfully replay the the log and complete all outstanding
meta
data requests it assumes the file system is clean, if not it simply
won't
mount.
If at any time a live XFS file system detects and error XFS will
immediately halt
all IO to the device. It is then up to the administrator to do manual
checks and
or repairs.
-Russell
|