linux-btrfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Goffredo Baroncelli <kreijack@inwind.it>
To: Anand Jain <anand.jain@oracle.com>, linux-btrfs@vger.kernel.org
Subject: Re: [survey] sysfs layout for btrfs
Date: Tue, 18 Aug 2015 14:34:23 +0200	[thread overview]
Message-ID: <55D3264F.6020002@inwind.it> (raw)
In-Reply-To: <55CE7C78.1020509@oracle.com>

Hi
On 2015-08-15 01:40, Anand Jain wrote:
> Hello,
> 
> as of now btrfs sysfs does not include the attributes for the volume manager part in its sysfs layout, so its being developed and there are two types of layout here below, so I have a quick survey to know which will be preferred. contenders are:
> 1. FS and VM (volume manager) attributes[1] merged sysfs layout
> 2. FS and VM attributes separated sysfs layout.
> 
> These two choices differ whether the VM attributes are amalgamate with existing FS attributes or if VM attributes are put under a kobject named "pools/volumes" under /sys/fs/btrfs. More in the below example. which would highlight the trade off between these two.
> 
> Eg for #1 (above):
> The existing sysfs for btrfs, has the top kobject <fsid>
> 
>   /sys/fs/btrfs/<fsid> <-- holds FS attr, VM attr will be added here.
>   /sys/fs/btrfs/<fsid>/devices/<uuid [2] > <-- btrfs_devices attr here
>   /sys/fs/btrfs/<fsid>/devices/<uuid>/state
>   /sys/fs/btrfs/<fsid>/devices/<uuid>/offline
> 
> we won't be able to change the sysfs entries which is already there. However we could change the context in which they are created and destroyed that is, from mount and unmount, to device scan and module unload respectively. And so this will enable us to implement the # 1 (above).
> 
> Eg for #2 (above):
> For the 2nd choice, a new 'pools or volumes' kobject will be created under existing /sys/fs/btrfs/ which will hold the VM attributes. (however note that: there will be duplicate kobjects like <fsid> both under FS and VM in this choice #2).
> 
>  /sys/fs/btrfs/<fsid> <--- as is, will continue to hold fs attributes.
>  /sys/fs/btrfs/pools/<fsid>/ <-- will hold VM attributes
>  /sys/fs/btrfs/pools/<fsid>/devices/<sdx> <-- btrfs_devices attr here
>  /sys/fs/btrfs/pools/<fsid>/devices/<sdx>/state
>  /sys/fs/btrfs/pools/<fsid>/devices/<sdx>/offline
> 
> There is certainly a small trade-off between these two. Your comments / feedback are kindly appreciated.

If the info "VM attributes" are per filesystem, I don't see any reason to put these under another subdirectory. So for this I vote #1

However I find more interesting to discuss where put the <uuid> entries.
What confusing me is that under <fsid>/devices/ there are both a link to the devices, and a directory <uuid> with disk-information..

It could be possible to get rid of the <devX> link ? I know that the official answer is "the backward compatibility must be preserved"; but who use these link today ? 
Otherwise it could be possible to create another directory (called dev-uuid/ for example) where put the <dev-uuid> ? My rationale is: if there is a collection of homogeneous entries (like a list of disk-uuid or filesystem-uuid), this collection (and only this collection) must be put in a dedicated directory.

BR
G.Baroncelli

> 
> Thanks, Anand
> 
> [1] attributes will be of the btrfs_fs_devices structure. And few newly introduced attributes like 'state', to state the volume current state.
> 
> [2] note that we can not use <sdx> here since a link to the block device already exists with that name.
> -- 
> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 


-- 
gpg @keyserver.linux.it: Goffredo Baroncelli <kreijackATinwind.it>
Key fingerprint BBF5 1610 0B64 DAC6 5F7D  17B2 0EDA 9B37 8B82 E0B5

      parent reply	other threads:[~2015-08-18 12:34 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-08-14 23:40 [survey] sysfs layout for btrfs Anand Jain
2015-08-18  8:33 ` Gabriel de Perthuis
2015-08-18 12:34 ` Goffredo Baroncelli [this message]

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=55D3264F.6020002@inwind.it \
    --to=kreijack@inwind.it \
    --cc=anand.jain@oracle.com \
    --cc=linux-btrfs@vger.kernel.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
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).