[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: xfs file system - apparently for real men 8-)

On Wed, 05 Dec 2001, Bret Waldow wrote:
> Alexander Clouter wrote:
>>  As for the XFS.....GOOD MAN :) None of that 'girly-man' reiserfs
>>  crud :)
>> I'm running it too and everything is just peachy :)
> Ok, I'll bite.  What's bad about reiser, or what's good about xfs?
> Seriously, if there's something serious there - inquiring minds and
> all that...

Well, personally I use ext3, and I wouldn't let ReiserFS near anything
that I cared about. Heck, I even look suspiciously at the sources... ;)

Seriously, though, I really wouldn't advise ReiserFS to anyone, at least
not for a year or two. The reasons are:

1) static fsck tool /very/ poor[1]
2) bi-monthly "ReiserFS ate my filesystem" posts on linux-kernel
3) semi-regular "ReiserFS needs this patch to not die in situation X"
   posts on same.
4) "Just run reiserfsck --rebuild-tree to make sure it keeps working..."
5) Performances falls on the floor one the FS is > 90% full, on average.
6) Only journals meta-data.

It's got some *great* ideas in it, and it's really well supported by the
people who are developing it. I have a great deal of respect for them,
don't get me wrong.

I just think that they made a mistake building fsck second and I
(personally, and I am paranoid) like journaled data.

The stability of it has never been sufficient for me. It seems to be an
endless treadmill of fix after fix after fix, each removing an
occasional data corruption bug.

Heck, tail packing spend over a year where they had corruption of files
regularly. It /looks/ stable now, a year later, because no one has
reported a bug in the area for a few months -- but do you believe that?


[1]  Every single time I looked at it. I have not done so for, like,
     three months. If it's great now, someone correct me, please. :)

What is art but a way of seeing?
        -- Thomas Berger

Reply to: