mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: david@lang.hm
To: Matthew Garrett <mjg59@srcf.ucam.org>
Cc: Theodore Tso <tytso@mit.edu>, Sitsofe Wheeler <sitsofe@yahoo.com>,
	"Andreas T.Auer" <andreas.t.auer_lkml_73537@ursus.ath.cx>,
	Alberto Gonzalez <info@gnebu.es>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: Ext4 and the "30 second window of death"
Date: Thu, 2 Apr 2009 11:22:48 -0700 (PDT)	[thread overview]
Message-ID: <alpine.DEB.1.10.0904021116500.28893@asgard.lang.hm> (raw)
In-Reply-To: <20090401174336.GA14726@srcf.ucam.org>

On Wed, 1 Apr 2009, Matthew Garrett wrote:

>> The other subtlety comes if we add fsync() suppression to laptop mode
>> --- which is something that Bart Samwel is very interested in doing
>> and I talked to him at FOSDEM about this.  As Jeff Garzik recently
>> pointed out, however, if we let the system reorder writes across
>> fsync() boundaries, or if we combine two writes to the same block
>> separated by an fsync(), and the system crashes in the middle of
>> pushing all of these blocks out to the disk, we can end up trashing
>> the consistency guarantees of a database such as mysql or postgres.
>> It's a good point, but it only applies if we add fsync() suppression
>> to laptop mode --- which we haven't done yet.
>
> I've got absolutely no idea why anyone would want fsync() to stop
> meaning "Put my data on the disk please". laptop-mode isn't intended to
> reduce data integrity - it's intended to batch disk write-outs such that
> there's a lower risk of needing to perform further write-outs in future.
> It makes sense for applications which really desperately want
> information on disk to fsync() (for instance, saving a file in
> OpenOffice).
>
> laptop-mode is something that makes sense as a default behaviour under a
> lot of circumstances. Adding fsync() suppression means it's utterly
> impossible to use it in that way. An additional mode would be perfectly
> reasonable, as long as it's made clear that it's really a request for
> data to be discarded at some point. The current mode isn't.

this issue seems pretty straightforward to me

the apps do fsync (and similar) to the degree that they think their data 
is important (potentially with config options if they acknowlege that 
their data isn't _always_ that important)

the system allows the admin to override the application and say "I'm 
willing to loose up to X seconds of data for other benifits"

if this can work cleanly (with the ordering issue that was identified, 
which may involve having multiple versions of the metadata cached) it 
seems like a very clean interface.

David Lang


  parent reply	other threads:[~2009-04-02 18:23 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-03-29 10:24 Alberto Gonzalez
2009-03-31 12:25 ` Theodore Tso
2009-03-31 12:52   ` Alberto Gonzalez
2009-03-31 13:45     ` Theodore Tso
2009-03-31 14:45       ` Alberto Gonzalez
2009-04-01  0:04         ` Theodore Tso
2009-04-01  1:14           ` Alberto Gonzalez
2009-03-31 22:02       ` Alberto Gonzalez
2009-03-31 23:22         ` Andreas T.Auer
2009-04-01  1:25           ` Alberto Gonzalez
2009-04-01  1:50           ` Theodore Tso
2009-04-01  5:20             ` Sitsofe Wheeler
2009-04-01 15:12               ` Matthew Garrett
2009-04-01 17:35                 ` Theodore Tso
2009-04-01 17:43                   ` Matthew Garrett
2009-04-01 21:21                     ` Ray Lee
2009-04-01 21:26                       ` Matthew Garrett
2009-04-02 11:25                       ` Sitsofe Wheeler
2009-04-02 18:22                     ` david [this message]
2009-04-02 18:29                       ` Matthew Garrett
2009-04-02 18:44                         ` david
2009-04-02 20:07                           ` Ray Lee
2009-04-02 20:59                             ` Andreas T.Auer
2009-04-02 23:38                               ` Theodore Tso
2009-04-03  0:00                                 ` Matthew Garrett
2009-04-03  7:33                                 ` Pavel Machek
2009-04-03  8:14                                 ` Andreas T.Auer
2009-04-02 22:36                           ` Bron Gondwana
2009-04-02 23:46                           ` Matthew Garrett
2009-04-03  0:55                             ` david
2009-04-03  1:06                               ` Matthew Garrett
2009-04-03  1:16                                 ` david
2009-04-03  1:19                                   ` Matthew Garrett
2009-04-03  1:24                                     ` david
2009-04-03  1:36                                       ` Matthew Garrett
2009-04-03  3:08                                         ` david
2009-04-03 13:42                                           ` Matthew Garrett
2009-04-03  4:54                                         ` Theodore Tso
2009-04-03 11:09                                           ` Sitsofe Wheeler
2009-04-03 13:07                                           ` Alberto Gonzalez
2009-04-03 13:45                                           ` Matthew Garrett
2009-04-02 18:34                       ` Nick Piggin
2009-04-02 18:38                         ` Matthew Garrett
2009-04-02 18:56                           ` Nick Piggin
2009-04-02 23:47                             ` Matthew Garrett
2009-04-03  0:59                               ` david
2009-04-03  1:09                                 ` Matthew Garrett
2009-04-03  1:17                                   ` david
2009-04-03  1:22                                     ` Matthew Garrett
2009-04-03  2:22                             ` Ric Wheeler
2009-04-02 21:47                         ` david
2009-04-06 21:32                     ` supporting laptops fs-semantic changes (was Re: Ext4 and the "30 second window of death") Linda Walsh
2009-04-02 11:37                   ` Ext4 and the "30 second window of death" Sitsofe Wheeler
2009-04-01  8:51             ` Andreas T.Auer
2009-04-03  7:13   ` Bojan Smojver
2009-04-05  4:07     ` Bojan Smojver
2009-04-05  4:51       ` Bojan Smojver
2009-04-05  5:41       ` Bojan Smojver
2009-04-05 17:27   ` Ed Tomlinson
2009-04-05 18:13 Tomasz Chmielewski

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=alpine.DEB.1.10.0904021116500.28893@asgard.lang.hm \
    --to=david@lang.hm \
    --cc=andreas.t.auer_lkml_73537@ursus.ath.cx \
    --cc=info@gnebu.es \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mjg59@srcf.ucam.org \
    --cc=sitsofe@yahoo.com \
    --cc=tytso@mit.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®