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

Re: The fsync issue

On Sat, 2010-11-27 at 07:59 +0100, Guillem Jover wrote:
> Hi Ben!
> On Fri, 2010-11-26 at 13:31:20 +0000, Ben Hutchings wrote:
> > Just got this from Christoph Helwig:
> > 
> > 13:23 < hch> bwh: if you guys are interested in helping dpkg review and ack the 
> >              per-fs sync ioctl path that sage weil sent out a couple of weeks 
> >              ago
> > 13:24 < hch> bwh: and report the ext4 fsync issues to the list, I know ext4 
> >              fsync isn't stellar, but the numbers sounds so bad that there must 
> >              be a bug somewhere
> > The patch referred to is in
> > <http://thread.gmane.org/gmane.linux.file-systems/44628>.
> Unfortunately that patch does not seem much appealing, it's Linux only,
> not even in mainline, and it would need for dpkg to track on which file
> system each file is located and issue such ioctl once per file system.

You don't need to tell me this.

> I'd rather not complicate the dpkg source code even more for something
> that seems to me to be a bug or missfeature in the file system. More so
> when there's a clear fix (nodelalloc) that solves both the performance
> and data safety issues in general.

But that 'clear fix' is bad for performance in general, as delayed
allocation reduces fragmentation.  Please talk to upstream about the bad
fsync() performance.


Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.

Attachment: signature.asc
Description: This is a digitally signed message part

Reply to: