There's a slightly newer version of the lilo raid1 patches available at
http://www.elliott-turbo.com/lilo/ which takes into account some of the
isues you've mentioned.
Rich B
----- Original Message -----
From: "Martin Bene" <[EMAIL PROTECTED]>
To: "Neil Brown" <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Thursday, June 08, 2000 2:20 AM
Subject: AW: Patches for Linux Software RAID in 2.4.0-test1
> Hi Neil,
>
> > 1/ grab my patches, read them, and comment/complain
> > 2/ grab my patches and test them
> > 3/ send me other patches for possible inclusion in my set
> > 4/ send me bug reports (cc to linux-raid) and I will investigate
> > where I can
>
> Just a tiny one: there's a long-standing problem with creating arrays
using
> the failed-disk raidtab directive; you end up with an array where nr_disks
> is too high by one. This will cause lilo-raid1 to fail miserably BTW, no
> other adverse effects are currently known.
>
> the small patch at
> ftp://ftp.sime.com/pub/linux/raidtools-19990824-0.90.mabene.gz should fix
> this.
>
> Another thing: when looking over the raid1 lilo patches I found that
> handling of failed / hotadded disks in current raid is rather strange with
> regard to nr_disks:
>
> * Healty mirror: nr_disks = 2, 1 = active; 2 = active
> * first disk fails: nr_disks = 1, 1 = failed; 2 = active
> * add spare: nr_disks = 2, 1 = failed; 2 = active; 3=spare/sync
> running
> * resync finished: nr_disks = 2, 1 = new active disk; 2 = active.
>
>
> The fact that nr_disks can get lower than the number of a disk actice in
the
> array causes no end of trouble for external programs trying to work with
> raid (like patched lilo): it's now impossible to get info on the active
disk
> since number of active disk > nr_disks and thus get_info failes.
>
> I'd think correct handlink should be:
> * Healty mirror: nr_disks = 2, 1 = active; 2 = active
> * first disk fails: nr_disks = 2, 1 = failed; 2 = active
> * add spare: nr_disks = 3, 1 = failed; 2 = active; 3=spare/sync
> running
> * resync finished: nr_disks = 2, 1 = new active disk; 2 = active.
>
> I've repeatedly failed to get any reaction whatsoever (is it correct the
way
> it is? would a patch be accepted? should the right behaviour be
different?)
> out of Ingo, so I'm bothering you instead :-)
>
> Bye, Martin
>
> "you have moved your mouse, please reboot to make this change take effect"
> --------------------------------------------------
> Martin Bene vox: +43-316-813824
> simon media fax: +43-316-813824-6
> Andreas-Hofer-Platz 9 e-mail: [EMAIL PROTECTED]
> 8010 Graz, Austria
> --------------------------------------------------
> finger [EMAIL PROTECTED] for PGP public key
>
