From: Josef Bacik <josef@toxicpanda.com>
To: "J. Bruce Fields" <bfields@fieldses.org>, NeilBrown <neilb@suse.de>
Cc: Christoph Hellwig <hch@infradead.org>,
Chuck Lever <chuck.lever@oracle.com>, Chris Mason <clm@fb.com>,
David Sterba <dsterba@suse.com>,
Alexander Viro <viro@zeniv.linux.org.uk>,
linux-fsdevel@vger.kernel.org, linux-nfs@vger.kernel.org,
linux-btrfs@vger.kernel.org
Subject: Re: [PATCH/RFC 00/11] expose btrfs subvols in mount table correctly
Date: Wed, 28 Jul 2021 17:30:04 -0400 [thread overview]
Message-ID: <e75ccfd2-e09f-99b3-b132-3bd69f3c734c@toxicpanda.com> (raw)
In-Reply-To: <20210728193536.GD3152@fieldses.org>
On 7/28/21 3:35 PM, J. Bruce Fields wrote:
> I'm still stuck trying to understand why subvolumes can't get their own
> superblocks:
>
> - Why are the performance issues Josef raises unsurmountable?
> And why are they unique to btrfs? (Surely there other cases
> where people need hundreds or thousands of superblocks?)
>
I don't think anybody has that many file systems. For btrfs it's a single file
system. Think of syncfs, it's going to walk through all of the super blocks on
the system calling ->sync_fs on each subvol superblock. Now this isn't a huge
deal, we could just have some flag that says "I'm not real" or even just have
anonymous superblocks that don't get added to the global super_blocks list, and
that would address my main pain points.
The second part is inode reclaim. Again this particular problem could be
avoided if we had an anonymous superblock that wasn't actually used, but the
inode lru is per superblock. Now with reclaim instead of walking all the
inodes, you're walking a bunch of super blocks and then walking the list of
inodes within those super blocks. You're burning CPU cycles because now instead
of getting big chunks of inodes to dispose, it's spread out across many super
blocks.
The other weird thing is the way we apply pressure to shrinker systems. We
essentially say "try to evict X objects from your list", which means in this
case with lots of subvolumes we'd be evicting waaaaay more inodes than you were
before, likely impacting performance where you have workloads that have lots of
files open across many subvolumes (which is what FB does with it's containers).
If we want a anonymous superblock per subvolume then the only way it'll work is
if it's not actually tied into anything, and we still use the primary super
block for the whole file system. And if that's what we're going to do what's
the point of the super block exactly? This approach that Neil's come up with
seems like a reasonable solution to me. Christoph gets his separation and
/proc/self/mountinfo, and we avoid the scalability headache of a billion super
blocks. Thanks,
Josef
next prev parent reply other threads:[~2021-07-28 21:30 UTC|newest]
Thread overview: 122+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-07-27 22:37 [PATCH/RFC 00/11] expose btrfs subvols in mount table correctly NeilBrown
2021-07-27 22:37 ` [PATCH 07/11] exportfs: Allow filehandle lookup to cross internal mount points NeilBrown
2021-07-28 10:13 ` Amir Goldstein
2021-07-29 0:28 ` NeilBrown
2021-07-29 5:27 ` Amir Goldstein
2021-08-06 7:52 ` Miklos Szeredi
2021-08-06 8:08 ` Amir Goldstein
2021-08-06 8:18 ` Miklos Szeredi
2021-07-28 19:17 ` J. Bruce Fields
2021-07-28 22:25 ` NeilBrown
2021-07-27 22:37 ` [PATCH 04/11] VFS: export lookup_mnt() NeilBrown
2021-07-30 0:31 ` Al Viro
2021-07-30 5:33 ` NeilBrown
2021-07-27 22:37 ` [PATCH 01/11] VFS: show correct dev num in mountinfo NeilBrown
2021-07-30 0:25 ` Al Viro
2021-07-30 5:28 ` NeilBrown
2021-07-30 5:54 ` Miklos Szeredi
2021-07-30 6:13 ` NeilBrown
2021-07-30 7:18 ` Miklos Szeredi
2021-07-30 7:33 ` NeilBrown
2021-07-30 7:59 ` Miklos Szeredi
2021-08-02 4:18 ` A Third perspective on BTRFS nfsd subvol dev/inode number issues NeilBrown
2021-08-02 5:25 ` Al Viro
2021-08-02 5:40 ` NeilBrown
2021-08-02 7:54 ` Amir Goldstein
2021-08-02 13:53 ` Josef Bacik
2021-08-03 22:29 ` Qu Wenruo
2021-08-02 14:47 ` Frank Filz
2021-08-02 21:24 ` NeilBrown
2021-08-02 7:15 ` Martin Steigerwald
2021-08-02 21:40 ` NeilBrown
2021-08-02 12:39 ` J. Bruce Fields
2021-08-02 20:32 ` Patrick Goetz
2021-08-02 20:41 ` J. Bruce Fields
2021-08-02 21:10 ` NeilBrown
2021-08-02 21:50 ` J. Bruce Fields
2021-08-02 21:59 ` NeilBrown
2021-08-02 22:14 ` J. Bruce Fields
2021-08-02 22:36 ` NeilBrown
2021-08-03 0:15 ` J. Bruce Fields
2021-07-27 22:37 ` [PATCH 03/11] VFS: pass lookup_flags into follow_down() NeilBrown
2021-07-27 22:37 ` [PATCH 11/11] btrfs: use automount to bind-mount all subvol roots NeilBrown
2021-07-28 8:37 ` kernel test robot
2021-07-28 8:37 ` [RFC PATCH] btrfs: btrfs_mountpoint_expiry_timeout can be static kernel test robot
2021-07-28 13:12 ` [PATCH 11/11] btrfs: use automount to bind-mount all subvol roots Christian Brauner
2021-07-29 0:43 ` NeilBrown
2021-07-29 14:38 ` Christian Brauner
2021-07-31 6:25 ` [btrfs] 5874902268: xfstests.btrfs.202.fail kernel test robot
2021-07-27 22:37 ` [PATCH 06/11] nfsd: include a vfsmount in struct svc_fh NeilBrown
2021-07-27 22:37 ` [PATCH 10/11] btrfs: introduce mapping function from location to inum NeilBrown
2021-07-27 22:37 ` [PATCH 02/11] VFS: allow d_automount to create in-place bind-mount NeilBrown
2021-07-27 22:37 ` [PATCH 09/11] nfsd: Allow filehandle lookup to cross internal mount points NeilBrown
2021-07-28 19:15 ` J. Bruce Fields
2021-07-28 22:29 ` NeilBrown
2021-07-30 0:42 ` Al Viro
2021-07-30 5:43 ` NeilBrown
2021-07-27 22:37 ` [PATCH 08/11] nfsd: change get_parent_attributes() to nfsd_get_mounted_on() NeilBrown
2021-07-27 22:37 ` [PATCH 05/11] VFS: new function: mount_is_internal() NeilBrown
2021-07-28 2:16 ` Al Viro
2021-07-28 3:32 ` NeilBrown
2021-07-30 0:34 ` Al Viro
2021-07-28 2:19 ` [PATCH/RFC 00/11] expose btrfs subvols in mount table correctly Al Viro
2021-07-28 4:58 ` Wang Yugui
2021-07-28 6:04 ` Wang Yugui
2021-07-28 7:01 ` NeilBrown
2021-07-28 12:26 ` Neal Gompa
2021-07-28 19:14 ` J. Bruce Fields
2021-07-29 1:29 ` Zygo Blaxell
2021-07-29 1:43 ` NeilBrown
2021-07-29 23:20 ` Zygo Blaxell
2021-07-28 22:50 ` NeilBrown
2021-07-29 2:37 ` Zygo Blaxell
2021-07-29 3:36 ` NeilBrown
2021-07-29 23:20 ` Zygo Blaxell
2021-07-30 2:36 ` NeilBrown
2021-07-30 5:25 ` Qu Wenruo
2021-07-30 5:31 ` Qu Wenruo
2021-07-30 5:53 ` Amir Goldstein
2021-07-30 6:00 ` NeilBrown
2021-07-30 6:09 ` Qu Wenruo
2021-07-30 5:58 ` NeilBrown
2021-07-30 6:23 ` Qu Wenruo
2021-07-30 6:53 ` NeilBrown
2021-07-30 7:09 ` Qu Wenruo
2021-07-30 18:15 ` Zygo Blaxell
2021-07-30 15:17 ` J. Bruce Fields
2021-07-30 15:48 ` Josef Bacik
2021-07-30 16:25 ` Forza
2021-07-30 17:43 ` Zygo Blaxell
2021-07-30 5:28 ` Amir Goldstein
2021-07-28 13:43 ` g.btrfs
2021-07-29 1:39 ` NeilBrown
2021-07-29 9:28 ` Graham Cobb
2021-07-28 7:06 ` NeilBrown
2021-07-28 9:36 ` Wang Yugui
2021-07-28 19:35 ` J. Bruce Fields
2021-07-28 21:30 ` Josef Bacik [this message]
2021-07-30 0:13 ` Al Viro
2021-07-30 6:08 ` NeilBrown
2021-08-13 1:45 ` [PATCH] VFS/BTRFS/NFSD: provide more unique inode number for btrfs export NeilBrown
2021-08-13 14:55 ` Josef Bacik
2021-08-15 7:39 ` Goffredo Baroncelli
2021-08-15 19:35 ` Roman Mamedov
2021-08-15 21:03 ` Goffredo Baroncelli
2021-08-15 21:53 ` NeilBrown
2021-08-17 19:34 ` Goffredo Baroncelli
2021-08-17 21:39 ` NeilBrown
2021-08-18 17:24 ` Goffredo Baroncelli
2021-08-15 22:17 ` NeilBrown
2021-08-19 8:01 ` Amir Goldstein
2021-08-20 3:21 ` NeilBrown
2021-08-20 6:23 ` Amir Goldstein
2021-08-23 4:05 ` [PATCH v2] BTRFS/NFSD: " NeilBrown
2021-08-18 14:54 ` [PATCH] VFS/BTRFS/NFSD: " Wang Yugui
2021-08-18 21:46 ` NeilBrown
2021-08-19 2:19 ` Zygo Blaxell
2021-08-20 2:54 ` NeilBrown
2021-08-22 19:29 ` Zygo Blaxell
2021-08-23 5:51 ` NeilBrown
2021-08-23 23:22 ` NeilBrown
2021-08-25 2:06 ` Zygo Blaxell
2021-08-23 0:57 ` Wang Yugui
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=e75ccfd2-e09f-99b3-b132-3bd69f3c734c@toxicpanda.com \
--to=josef@toxicpanda.com \
--cc=bfields@fieldses.org \
--cc=chuck.lever@oracle.com \
--cc=clm@fb.com \
--cc=dsterba@suse.com \
--cc=hch@infradead.org \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=neilb@suse.de \
--cc=viro@zeniv.linux.org.uk \
/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;
as well as URLs for NNTP newsgroup(s).