quota deadlock on 6.1-RC1

Pawel Jakub Dawidek pjd at FreeBSD.org
Fri May 5 08:07:29 UTC 2006

On Thu, May 04, 2006 at 01:54:58AM -0400, Mike Jakubik wrote:
+> Kris Kennaway wrote:
+> >On Thu, May 04, 2006 at 12:41:57AM -0400, Mike Jakubik wrote:
+> >>Then why utilize a known non-functional technology?
+> >>    
+> >
+> >Because again, the benefits have been judged by the decision-makers
+> >and found to outweigh the drawbacks.  Perhaps that's just a difficult
+> >concept for some people to understand if they're used to thinking of
+> >everything in binary terms.
+> >  
+> Yes, i am sorry, but i fail to understand why i would want to use something that i know does not work correctly. I think there are quite a few of those "drawbacks" that are 
+> pissed off.

I'm using bgfsck very often and I didn't have problems with it.
Those hangs aren't so easy to trigger in everyday use.

There are serval committers, including me and Kris who work on those
hangs very hard for more than two week now. The problem is that VFS is
very complex and there are many subtle bugs.

We found few more problems with snapshots, which weren't reported by
users, because of our extensive testing. Fixing one bug, uncovers
another one, etc., but as I said those bug don't touch every user and
don't make UFS to hang always making FreeBSD unusable.

Some of those bugs are maybe quite easy to fix, so the only risk here
are latent bugs the fix can uncover, but some of them need a lot of work
to fix properly and be sure nothing else will break.
That's why fixing those bugs must include extensive testing on many
different work-loads. We can't just commit those and hope for the best.

The point here is that FreeBSD is as good as their developers are
professional and responsible and belive me, committing quick hacks to
fix those issues can break 6.1-RELEASE for much more users, who will
then send mails to freebsd-stable@ saying "Is FreeBSD a professional
operating system or a joke? How can they ship a release with broken,
untested code?". Do you think that answering "We had two users who
insisted on fixing those bugs just before release, blame them!" would
satisfy them?

It isn't good to release a software with known, documented bugs, but its
better than shipping an untested software with god-one-knows unknown

Pawel Jakub Dawidek                       http://www.wheel.pl
pjd at FreeBSD.org                           http://www.FreeBSD.org
FreeBSD committer                         Am I Evil? Yes, I Am!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20060505/1eca45c3/attachment.pgp

More information about the freebsd-stable mailing list