From: NeilBrown <neilb@suse.de>
To: Andy Smith <andy@strugglers.net>
Cc: linux-raid@vger.kernel.org
Subject: Re: grub-probe: "found two disks with the index 3 for RAID md2" after changing superblock 0.9 to 1.0
Date: Thu, 18 Sep 2014 16:49:00 +1000 [thread overview]
Message-ID: <20140918164900.3d526d69@notabene.brown> (raw)
In-Reply-To: <20140916040023.GG27492@bitfolk.com>
[-- Attachment #1: Type: text/plain, Size: 1954 bytes --]
On Tue, 16 Sep 2014 04:00:24 +0000 Andy Smith <andy@strugglers.net> wrote:
> Hi Neil,
>
> On Tue, Sep 16, 2014 at 07:54:58AM +1000, NeilBrown wrote:
> > On Mon, 15 Sep 2014 17:12:52 +0000 Andy Smith <andy@strugglers.net> wrote:
> > > On Mon, Sep 15, 2014 at 05:00:37PM +1000, NeilBrown wrote:
> > > > mdadm --zero-super --metadata=0.90 /dev/sda3
> > > >
> > > > will remove any 0.90 metadata.
> > >
> > > Trying that whole the array is running:
> > >
> > > mdadm: Couldn't open /dev/sda3 for write - not zeroing
> > >
> > > This needs to be done with the array stopped, presumably?
> >
> > It's preferred, but if you are really confident (and I think you are), then
> > adding --force is justified and should make it work.
>
> Yep, that worked, thank you! No more complaints from grub-probe.
>
> Do you consider it a bug that mdadm does not zero the old metadata
> when upgrading it to a new version?
No, but it might be excessive caution. I think it was a deliberate decision
to make it (relatively) easy to revert if anything when wrong.
mdadm always prefers the more recent metadata, so the old metadata is
effectively invisible to mdadm. Not to grub though.
>
> Or do you consider it a documentation issue to explain that it may
> be necessary to manually do it?
Maybe. Though I really don't want to impose that sort of burden on users.
Things should "just work". Maybe I should bite the bullet and just erase
the old metadata.... if I do it *after* the array is assembled it should be
safe.
The code might be a bit clumsy but that isn't the end of the world...
I've made a not on my to-do list :-)
Thanks,
NeilBrown
>
> I think I would've liked grub-probe at least in verbose mode to say
> which version of metadata it was reading from where. That would've
> been immediately enlightening, so I'll see if they would accept a
> patch for that…
>
> Cheers,
> Andy
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 828 bytes --]
prev parent reply other threads:[~2014-09-18 6:49 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-09-15 6:47 grub-probe: "found two disks with the index 3 for RAID md2" after changing superblock 0.9 to 1.0 Andy Smith
2014-09-15 7:00 ` NeilBrown
2014-09-15 17:12 ` Andy Smith
2014-09-15 21:54 ` NeilBrown
2014-09-16 4:00 ` Andy Smith
2014-09-18 6:49 ` NeilBrown [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=20140918164900.3d526d69@notabene.brown \
--to=neilb@suse.de \
--cc=andy@strugglers.net \
--cc=linux-raid@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).