nfs and ext2 option for maximum relyability
alanr at suse.com
Fri Jul 28 13:14:51 MDT 2000
Jeremy Hansen wrote:
> I have been using ext3 on several machines as well as a lot of other
> people it seems (www.rpmfind.net) and stability definitely does seem to be
> the issue. There are things that are mainly goals for ext3 that it
> doesn't do yet, but from what I understand, these goals do not really
> affect its stability, but rather its performance.
> With ext3 you will not have to worry about quotas, nfs weirdness and the
> TLB IPI smp issues which still seem to plague reiserfs. Also ext3 is
> applied on top of ext2. You don't have to wipe out your filesystem to use
> ext3 and you can easily convert back to ext2.
I don't care about quotas (but certainly others might), but I'm not thrilled
about using the user-space NFS daemon. I don't think I'd eard of the TLB IPI
smp issues. Would you mind elaborating on that?
> I'm running it on things like my workstation, laptop. I don't even bother
> to shutdown anymore, heh.
> Steven Tweedie seems to believe that the current available version is very
> stable. Ext3 just seems like the right thing to do if you're looking for
> alternative. Decision like this I think go beyond just
> functionality. Hans Reiser seems to have pissed of a few key people in
> the kernel realm. Steven seems to work well with other developers.
Yes to both of these two last statements. Sadly to the first, and happily to
Nevertheless, there are also a lot of people giving ReiserFS a thorough testing,
and for many people it is working quite well.
But, if you need quotas they're not going to be there for a while. The knfsd
issues are being resolved, but aren't quite there yet.
And the REAL POINT is:
Of all people, the people on this list should be giving one or more of the 5
available journalling filesystems a thorough going over and beating them up
because we REALLY NEED THEM. And, in the Linux world that means we really need
to help with the testing. Even if you aren't ready to put them into production,
please help one of these teams test their code.
-- Alan Robertson
alanr at suse.com
More information about the Linux-HA