From kde-devel Mon Apr 21 16:12:28 2008 From: "Andreas Hartmetz" Date: Mon, 21 Apr 2008 16:12:28 +0000 To: kde-devel Subject: Re: fsync() madness Message-Id: X-MARC-Message: https://marc.info/?l=kde-devel&m=120879444220786 MIME-Version: 1 Content-Type: multipart/mixed; boundary="--===============0004070585==" --===============0004070585== Content-Type: multipart/alternative; boundary="----=_Part_3122_11961239.1208794348952" ------=_Part_3122_11961239.1208794348952 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline 2008/4/21, Luciano Montanaro : > > On Monday 21 April 2008 17:25:47 Gary Greene wrote: > > On Monday 21 April 2008 8:15:51 am Sami Liedes wrote: > > > On Mon, Apr 21, 2008 at 02:11:40PM +0200, Lubos Lunak wrote: > > > > On Sunday 20 of April 2008, Sami Liedes wrote: > > > > > I had missed that post. Still, no analysis of the performance hit > > > > > there, and I think the attitude of "no data loss at all allowed at > > > > > any power loss, implement at any cost to performance" is > misguided. > > > > > > > > Tell that to XFS developers and their users. Anyway, where's your > > > > patch? > > > > > > The patch is simple and not very fine grained, but effective and > > > shouldn't break anything unless a power loss happens. Attached. > > > > > > Sami > > > > Again you are not taking into account XFS. How many times must we > iterate > > over this.... _If you don't have the code check which FS this is on and > > PROPERLY deal with this, you will kill users data._ > > > Technically, it would be XFS that would do that. > > Alternatively, the user has made a choice that he *could* regret. Or he > could > be lucky, and fail to experience any data loss. > > So do we want to force a penalty for every user out there just to spare > some > potential grief to the tiny minority using an XFS filesystem? Hey, I use XFS and I don't feel like a tiny minority - it is one of the better filesystems :) We should also consider that other programs out there are not so zealous in > protecting user data... so if this were a real issue, people would steer > away > from the filesystem. Right. I use XFS and the frequent "hangs" on disk I/O that I get are a PITA while using the desktop. I don't care about config files *that* much... what about a "reliable flag" or something, or even explicit sync? Another approach is to hold off sync calls for some milliseconds to combine several calls. Luciano > > > ------=_Part_3122_11961239.1208794348952 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline

2008/4/21, Luciano Montanaro <mikelima@cirulla.net>:
On Monday 21 April 2008 17:25:47 Gary Greene wrote:
> On Monday 21 April 2008 8:15:51 am Sami Liedes wrote:
> > On Mon, Apr 21, 2008 at 02:11:40PM +0200, Lubos Lunak wrote:
> > > On Sunday 20 of April 2008, Sami Liedes wrote:
> > > > I had missed that post. Still, no analysis of the performance hit
> > > > there, and I think the attitude of "no data loss at all allowed at
> > > > any power loss, implement at any cost to performance" is misguided.
> > >
> > >  Tell that to XFS developers and their users. Anyway, where's your
> > > patch?
> >
> > The patch is simple and not very fine grained, but effective and
> > shouldn't break anything unless a power loss happens. Attached.
> >
> >     Sami
>
> Again you are not taking into account XFS. How many times must we iterate
> over this.... _If you don't have the code check which FS this is on and
> PROPERLY deal with this, you will kill users data._


Technically, it would be XFS that would do that.

Alternatively, the user has made a choice that he *could* regret. Or he could
be lucky, and fail to experience any data loss.

So do we want to force a penalty for every user out there just to spare some
potential grief to the tiny minority using an XFS filesystem?

Hey, I use XFS and I don't feel like a tiny minority - it is one of the better filesystems :) 

We should also consider that other programs out there are not so zealous in
protecting user data... so if this were a real issue, people would steer away
from the filesystem.

Right. I use XFS and the frequent "hangs" on disk I/O that I get are a PITA while using the desktop. I don't care about config files *that* much... what about a "reliable flag" or something, or even explicit sync?
Another approach is to hold off sync calls for some milliseconds to combine several calls.

Luciano


------=_Part_3122_11961239.1208794348952-- --===============0004070585== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline >> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe << --===============0004070585==--