linux-raid.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Mariusz Tkaczyk <mariusz.tkaczyk@linux.intel.com>
To: Stefan Fleischmann <sfle@kth.se>
Cc: Jes Sorensen <jes@trained-monkey.org>, linux-raid@vger.kernel.org
Subject: Re: [PATCH v2 0/6] mdadm: POSIX portable naming rules
Date: Wed, 3 Jan 2024 09:53:23 +0100	[thread overview]
Message-ID: <20240103095323.00002d36@linux.intel.com> (raw)
In-Reply-To: <20240101133255.520a6149@vinyamar>

On Mon, 1 Jan 2024 13:32:55 +0100
Stefan Fleischmann <sfle@kth.se> wrote:

> On 6/1/23 03:27, Mariusz Tkaczyk wrote:
> > Hi Jes,
> > To avoid problem with udev and VROC UEFI driver I developed stronger
> > naming policy, basing on POSIX portable names standard. Verification is
> > added for names and config entries. In case of an issue, user can update
> > name to make it compatible (for IMSM and native).
> > 
> > The changes here may cause /dev/md/ link will be different than before
> > mdadm update. To make any of that happen user need to use unusual naming
> > convention, like:
> > - special, non standard signs like, $,%,&,* or space.
> > - '-' used as first sign.
> > - locals.
> > 
> > Note: I didn't analyze configurations with "names=1". If name cannot be
> > determined mdadm should fallback to default numbered dev creation.
> > 
> > If you are planning release soon then feel free to merge it after the
> > release. It is a potential regression point.
> > 
> > It is a new version of [1] but it is strongly rebuild. Here is a list
> > of changes:
> > 1. negative and positive test scenarios for both create and config
> >    entries are added.
> > 2. Save parsed parameters in dedicated structs. It is a way to control
> >    what is parsed, assuming that we should use dedicated set_* function.
> > 3. Verification for config entries is added.
> > 5. Improved error logging for names:
> >    - during creation, these messages are errors, printed to stderr.
> >    - for config entries, messages are just a warnings printed to stdout.
> > 6. Error messages reworked.
> > 7. Updates in manual.
> > 
> > [1]
> > https://lore.kernel.org/linux-raid/20221221115019.26276-1-mariusz.tkaczyk@linux.intel.com/
> > 
> > Mariusz Tkaczyk (6):
> >   tests: create names_template
> >   tests: create 00confnames
> >   mdadm: set ident.devname if applicable
> >   mdadm: refactor ident->name handling
> >   mdadm: define ident_set_devname()
> >   mdadm: Follow POSIX Portable Character Set  
> 
> 
> Hi,
> 
> it seems to me that the behavior after this change is inconsistent. The
> default naming scheme for a created device is still <HOSTNAME>:<ID>. When
> I create a new MD device with:
> 
>  # mdadm --version
>  mdadm - v4.2 - 2021-12-30 - Debian 4.2+20231121-1
>  # mdadm --create /dev/md32 -l 1 -n 2 /dev/loop2 /dev/loop3
>  [...]
>  # mdadm --detail /dev/md32
>   /dev/md32:
>            Version : 1.2
>      Creation Time : Mon Jan  1 12:33:05 2024
>   [...]
>               Name : gondolin:32  (local to host gondolin)
>               UUID : d84b8e26:3675a722:61869b83:670699da
> 
> The name set automatically by mdadm is not allowed by the new naming
> rules, and mdadm complains about that later on. So for consistency one
> would either have to include : in the allowed character list or change
> the default naming scheme. Or am I missing something here?
> 
> Happy new year!
> Stefan

Hi Stefan,
Yeah, I noticed this problem recently but I do not fixed it yet. For sure it
will be fixed before release.
It does not prompt about the name in metadata but about name entry in config
file. I analyzed the usage on name from config and it seems to be something
which can be dropped but I will reconsider it deeper later.

Could you please test if removing "name=hostanme:id" from ARRAYLINE in
mdadm.conf fixes issue for you?

Thanks,
Mariusz

  reply	other threads:[~2024-01-03  8:55 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-01-01 12:32 [PATCH v2 0/6] mdadm: POSIX portable naming rules Stefan Fleischmann
2024-01-03  8:53 ` Mariusz Tkaczyk [this message]
  -- strict thread matches above, loose matches on Subject: below --
2023-06-01  7:27 Mariusz Tkaczyk
2023-06-01  8:57 ` Paul Menzel
2023-06-01  9:52   ` Mariusz Tkaczyk
2023-10-26 21:31 ` Jes Sorensen

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=20240103095323.00002d36@linux.intel.com \
    --to=mariusz.tkaczyk@linux.intel.com \
    --cc=jes@trained-monkey.org \
    --cc=linux-raid@vger.kernel.org \
    --cc=sfle@kth.se \
    /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).