Re: Creating RAID 1+0 with LVM
From: Frank Cusack (fcusack_at_fcusack.com)
Date: 08/02/05
- Next message: Jamie Beverly: "Re: Locale Issue"
- Previous message: Eric Enright: "Re: Locale Issue"
- In reply to: Darren Dunham: "Re: Creating RAID 1+0 with LVM"
- Next in thread: Gandalf: "Re: Creating RAID 1+0 with LVM"
- Reply: Gandalf: "Re: Creating RAID 1+0 with LVM"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Mon, 01 Aug 2005 19:35:32 -0700
On Tue, 02 Aug 2005 00:12:42 GMT Darren Dunham <ddunham@redwood.taos.com> wrote:
> Neil Truby <neil.truby@ardenta.com> wrote:
>>>> Obviously I'm missing something ...
>>>
>>> While the configuration syntax is 0+1, LVM actually creates 1+0 RAIDs.
>>> But don't believe me; you can easily test this.
>>>
>>> -frank
>
>> I do believe you ... but how can I test it? :-)
>
> Pull a disk. Watch that writes still hit both sides of the mirror where
> it remains redundant. Reconnect the disk and see that the resync is
> only for the slice with that disk, not the others.
I would have suggested, pull two drives, one on each side of the
mirror. Of course the two drives must not be the two submirrors of
the same part of the stripe. Observe that the md does not fail. This
proves 1+0 behavior.
eg
stripe 1 = drives 1a, 1b, 1c
stripe 2 = drives 2a, 2b, 2c
md = mirror of stripe 1 + 2
pull drive 1a
pull drive 2b
If this were really 0+1, pulling drive 1a would fail half of the
mirror (drives 1b and 1c would no longer be used at all). Pulling
drive 2b would then fail the other half and the md would fail. But in
1+0 it would continue chugging along.
-frank
- Next message: Jamie Beverly: "Re: Locale Issue"
- Previous message: Eric Enright: "Re: Locale Issue"
- In reply to: Darren Dunham: "Re: Creating RAID 1+0 with LVM"
- Next in thread: Gandalf: "Re: Creating RAID 1+0 with LVM"
- Reply: Gandalf: "Re: Creating RAID 1+0 with LVM"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
