Re: Journaling patches

[Andrew Doran <ad%netbsd.org@localhost>]

Hi Simon,

On Tue, Apr 22, 2008 at 11:32:09PM +1000, Simon Burge wrote:

> Unfortunately none of the outstanding issues have been solved yet with
> the journaling support.  However, I'd like to commit the patches to the
> source tree in the new few days - this will hopefully make it easier for
> more people to test and look at these issues.
> 
> In summary, the issues are:
> 
>  - Can't log the root filesystem (or any FS if /dev isn't an ffs).
>  - Log must currently live at end of partition.
>  - Some heavy workload performance issues on SMP machines.
>  - Some stability issues on SMP machines.
> 
> There is one other behavioural change.  You now need to use the "-f"
> option to mount a dirty filesystem.  I'm not sure of the original reason
> for this, and am not sure if this behaviour is desirable or not.  Any
> comments?
> 
> A patch against today's -current is at:
> 
>    http://www.NetBSD.org/~simonb/wapbl-current-20080422.diff

Some time back you, Greg and I briefly discussed moving the journalling
stuff towards a producer-consumer model where there is a kthread pushing
out journal updates to the disk. That might not be the best model. With the
patch as-is, updating threads take care of syncing the journal directly.

In the tests that I've done it works great as-is, but in a multiuser
environment it doesn't perform very well because the flushing serializes
everything. I'm concerned that if we commit this before 5.0, we could be
setting negative expectations of the feature. How do you feel about adding
a printf to ffs_mount() that indicates it's an experimental feature, until
we have fixed the concurrency issue?

Thanks,
Andrew