Git Mailing List Archive on lore.kernel.org
 help / color / Atom feed
From: Theodore Ts'o <tytso@mit.edu>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>,
	"Junio C Hamano" <gitster@pobox.com>,
	"Christoph Hellwig" <hch@lst.de>,
	"Git Mailing List" <git@vger.kernel.org>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>
Subject: Re: [PATCH] enable core.fsyncObjectFiles by default
Date: Wed, 17 Jan 2018 18:52:20 -0500
Message-ID: <20180117235220.GD6948@thunk.org> (raw)
In-Reply-To: <CA+55aFzJ2QO0MH3vgbUd8X-dzg_65A-jKmEBMSVt8ST2bpmzSQ@mail.gmail.com>

On Wed, Jan 17, 2018 at 02:07:22PM -0800, Linus Torvalds wrote:
> 
> Now re-do the test while another process writes to a totally unrelated
> a huge file (say, do a ISO file copy or something).
> 
> That was the thing that several filesystems get completely and
> horribly wrong. Generally _particularly_ the logging filesystems that
> don't even need the fsync, because they use a single log for
> everything (so fsync serializes all the writes, not just the writes to
> the one file it's fsync'ing).

Well, let's be fair; this is something *ext3* got wrong, and it was
the default file system back them.  All of the modern file systems now
do delayed allocation, which means that an fsync of one file doesn't
actually imply an fsync of another file.  Hence...

> The original git design was very much to write each object file
> without any syncing, because they don't matter since a new object file
> - by definition - isn't really reachable. Then sync before writing the
> index file or a new ref.

This isn't really safe any more.  Yes, there's a single log.  But
files which are subject to delayed allocation are in the page cache,
and just because you fsync the index file doesn't mean that the object
file is now written to disk.  It was true for ext3, but it's not true
for ext4, xfs, btrfs, etc.

The good news is that if you have another process downloading a huge
ISO image, the fsync of the index file won't force the ISO file to be
written out.  The bad news is that it won't force out the other git
object files, either.

Now, there is a potential downside of fsync'ing each object file, and
that is the cost of doing a CACHE FLUSH on a HDD is non-trivial, and
even on a SSD, it's not optimal to call CACHE FLUSH thousands of times
in a second.  So if you are creating thousands of tiny files, and you
fsync each one, each fsync(2) call is a serializing instruction, which
means it won't return until that one file is written to disk.  If you
are writing lots of small files, and you are using a HDD, you'll be
bottlenecked to around 30 files per second on a 5400 RPM HDD, and this
is true regardless of what file system you use, because the bottle
neck is the CACHE FLUSH operation, and how you organize the metadata
and how you do the block allocation, is largely lost in the noise
compared to the CACHE FLUSH command, which serializes everything.

There are solutions to this; you could simply not call fsync(2) a
thousand times, and instead write a pack file, and call fsync once on
the pack file.  That's probably the smartest approach.

You could also create a thousand threads, and call fsync(2) on those
thousand threads at roughly the same time.  Or you could use a
bleeding edge kernel with the latest AIO patch, and use the newly
added IOCB_CMD_FSYNC support.

But I'd simply recommend writing a pack and fsync'ing the pack,
instead of trying to write a gazillion object files.  (git-repack -A,
I'm looking at you....)

					- Ted

  parent reply index

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-01-17 18:48 Christoph Hellwig
2018-01-17 19:04 ` Junio C Hamano
2018-01-17 19:35   ` Christoph Hellwig
2018-01-17 20:05     ` Andreas Schwab
2018-01-17 19:37   ` Matthew Wilcox
2018-01-17 19:42     ` Christoph Hellwig
2018-01-17 21:44   ` Ævar Arnfjörð Bjarmason
2018-01-17 22:07     ` Linus Torvalds
2018-01-17 22:25       ` Linus Torvalds
2018-01-17 23:16       ` Ævar Arnfjörð Bjarmason
2018-01-17 23:42         ` Linus Torvalds
2018-01-17 23:52       ` Theodore Ts'o [this message]
2018-01-17 23:57         ` Linus Torvalds
2018-01-18 16:27           ` Christoph Hellwig
2018-01-19 19:08             ` Junio C Hamano
2018-01-20 22:14               ` Theodore Ts'o
2018-01-20 22:27                 ` Junio C Hamano
2018-01-22 15:09                   ` Ævar Arnfjörð Bjarmason
2018-01-22 18:09                     ` Theodore Ts'o
2018-01-23  0:47                       ` Jeff King
2018-01-23  5:45                         ` Theodore Ts'o
2018-01-23 16:17                           ` Jeff King
2018-01-23  0:25                     ` Jeff King
2018-01-21 21:32             ` Chris Mason
2020-09-17 11:06         ` Ævar Arnfjörð Bjarmason
2020-09-17 11:28           ` [RFC PATCH 0/2] should core.fsyncObjectFiles fsync the dir entry + docs Ævar Arnfjörð Bjarmason
2020-09-17 11:28           ` [RFC PATCH 1/2] sha1-file: fsync() loose dir entry when core.fsyncObjectFiles Ævar Arnfjörð Bjarmason
2020-09-17 13:16             ` Jeff King
2020-09-17 15:09               ` Christoph Hellwig
2020-09-17 14:09             ` Christoph Hellwig
2020-09-17 14:55               ` Jeff King
2020-09-17 14:56                 ` Christoph Hellwig
2020-09-17 15:37                   ` Junio C Hamano
2020-09-17 17:12                     ` Jeff King
2020-09-17 20:37                       ` Taylor Blau
2020-09-22 10:42               ` Ævar Arnfjörð Bjarmason
2020-09-17 20:21             ` Johannes Sixt
2020-09-22  8:24               ` Ævar Arnfjörð Bjarmason
2020-11-19 11:38                 ` Johannes Schindelin
2020-09-17 11:28           ` [RFC PATCH 2/2] core.fsyncObjectFiles: make the docs less flippant Ævar Arnfjörð Bjarmason
2020-09-17 14:12             ` Christoph Hellwig
2020-09-17 15:43             ` Junio C Hamano
2020-09-17 20:15               ` Johannes Sixt
2020-10-08  8:13               ` Johannes Schindelin
2020-10-08 15:57                 ` Ævar Arnfjörð Bjarmason
2020-10-08 18:53                   ` Junio C Hamano
2020-10-09 10:44                   ` Johannes Schindelin
2020-09-17 19:21             ` Marc Branchaud
2020-09-17 14:14           ` [PATCH] enable core.fsyncObjectFiles by default Christoph Hellwig
2020-09-17 15:30           ` Junio C Hamano
2018-01-17 20:55 ` Jeff King
2018-01-17 21:10   ` Christoph Hellwig
  -- strict thread matches above, loose matches on Subject: below --
2015-06-23 21:57 [PATCH] Enable " Stefan Beller
2015-06-23 22:21 ` Junio C Hamano
2015-06-23 23:29   ` Theodore Ts'o
2015-06-24  5:32     ` Junio C Hamano
2015-06-24 14:30       ` Theodore Ts'o
2015-06-24  1:07 ` Duy Nguyen
2015-06-24  3:37 ` Jeff King
2015-06-24  5:20   ` Junio C Hamano

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=20180117235220.GD6948@thunk.org \
    --to=tytso@mit.edu \
    --cc=avarab@gmail.com \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    --cc=hch@lst.de \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    /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

Git Mailing List Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/git/0 git/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 git git/ https://lore.kernel.org/git \
		git@vger.kernel.org
	public-inbox-index git

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.git


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git