Re: UFS2 Snapshots in 6.1-Beta4 - Confirmed Problems

On Thu, Mar 23, 2006 at 10:42:59AM -0800, John Kozubik wrote:
> 
> 
> On Thu, 23 Mar 2006, Kris Kennaway wrote:
> 
> > On Tue, Mar 21, 2006 at 01:36:44AM -0800, John Kozubik wrote:
> >
> > > First, I have confirmed that a filesystem with multiple snapshots that
> > > undergoes multiple, rapid deletions of files, will cause the system to
> > > hang.  I have witnessed this before, but had not confirmed it or
> > > documented it in a PR.  Now that I have confirmed this behavior, I have
> > > documented it in: kern/94769
> >
> > I ran a (completion of) your script in a loop for about 24 hours and
> > it didn't deadlock.  This may be because there was a second set of
> > fixes that was merged in to 6.x a day or two ago.  Are you able to
> > confirm whether they indeed fixed this problem?
> 
> What ratio of snapshots / rms did you run ?

3/3, as you suggested.

> I reproduced this easily on
> 6.0-RELEASE and 6.1-BETA4.  I am traveling right now and cannot test on
> anything newer.  Can you try it on one of the two above, confirm for
> yourself that it manifests itself, and then compare that to what you see
> on whatever newer code you have ?  If it is fixed, that's great.

Not right away, unfortunately.

> > > Second, kern/92292 is still a problem.  I have reproduced this error in
> > > 6.1-BETA4 (and have seen it happening since 5.1).  The (small) difference
> > > is that the cp process seems to stick in the flswai state instead of
> > > biowr.
> >
> > I'll test this one next.
> 
> 
> Great - thanks.  It's less urgent, as running a md-backed FS on a
> snapshotted FS is less common, and easily avoided.  The big fish are the
> rm problems (you can't avoid deleting things) and the "full filesystem"
> problems, since there is always a danger of that.

Yeah, the full filesystem bugs are a big pain, but as I mentioned
probably nothing can be done in time for 6.1 since it's a pretty big
job.

Kris