Question now that mandrake has all 4 fs' on board..
Granted that reiserfs can't hack bad blocks on the media, which of the
other ones will? ext2 can with a little help from 'badblocks', i'm not
sure if ext3 can hack a media error on its journal. Anyone know if XFS or
jfs can survive?
Frankly not having a straightforward way to handle bad blocks is broken
behaviour, and i was so disappointed with reiser when i knew it didn't.
I knew only about it when i had deployed several boxes into production..
Also, there's the issue of NFS. It took quite a while before one was able
to serve reiserfs partitions over NFS. Anyone know if there are similar
issues with JFS, ext3, and XFS? Has anyone tested this on the mandrake
team?
On Fri, 7 Sep 2001, Brian J. Murrell wrote:
> On Fri, Sep 07, 2001 at 03:38:51PM +0200, Gregoire Favre wrote:
> >
> > Is ext3 so good?
>
> Uhm, yeah! :-)
>
> > I have installed it on the root of my box, but found a
> > very annoying things: after 20 mounts (no: I haven't changed the
> > defaults...) the filesystem got checked (8 Gb with 7 Gb on it):
>
> You mean while the box was running? What do you mean by "checked"?
> fsck? How do you know it was being checked? I have been using ext3
> on 2.2 kernels for nearly a year now and not encountered anything like
> that.
>
> You sure somebody or something did not fire off an fsck?
>
> > The only advantage of ext3 I could see is that you could convert an ext2
> > to ext3 without reformating, but that's not that much...
>
> Not much? So what shall I do with my 40GB of data while I am
> reformatting filesystems just to get journalling? Why would I not
> prefer to just add a journal to existing filesystems?
>
> I won't even go into the argument of the "tried and _very_ tested"
> ext2 codebase vs. the "new and untested" (as much anyway) codebase of
> reiserfs. But let's not start a religeous war OK?
>
> b.
>
>
>
