RE: 2.2.10-14 i686 SMP: IDE RAID-5 array hangs on mount

From: Brendon B (brendon@brendon.com)
Date: Fri Feb 04 2000 - 22:50:22 EST


Okay, what about md v.90 in kernel >= 2.3.40 or 2.4? I checked 2.3.40 and
the old raid is in it. Where can i get a kernel that has new raid that is at
least 2.2.14 or better?

bb

-----Original Message-----
From: Peter Samuelson [mailto:peter@cadcamlab.org]
Sent: Wednesday, January 26, 2000 3:31 AM
To: Adam C Powell IV
Cc: linux-kernel@vger.rutgers.edu; debian-user@lists.debian.org
Subject: Re: 2.2.10-14 i686 SMP: IDE RAID-5 array hangs on mount

[Adam Powell <hazelsct@mit.edu>]
> Okay, but the current RAID in 2.2.14 doesn't work now (for SMP), and
> doesn't work right (if it's being replaced). I guess one could ask,
> how did this happen?

IIRC, Ingo was rewriting RAID during the Linux 2.1 cycle. Somehow it
didn't make it into the main tree before Linus declared the (first)
feature freeze. The new-style RAID is better than the old in all
respects, *but* it requires a new userspace toolset, so it is *not* in
any way a drop-in replacement.

After 2.2 came out (or maybe shortly before), some people started
pushing hard to get RAID 0.90 into the mainstream kernel. But by then
it was too late. You see, as bad as it looks to have old, broken RAID
in the kernel and new 'n' improved RAID in a patch somewhere, it looks
just as bad (from some people's perspective anyway) to force people to
upgrade their userspace tools when moving from (say) kernel 2.2.3 to
2.2.4.

Similar with knfsd and ISDN. When 2.2.0 shipped, much-improved
versions of both knfsd and ISDN were out there, but the diffs were too
large to drop them into a kernel that was supposed to be near-stable.
Happily, Alan decided that for 2.2.14, since the new knfsd does not
require userspace tool upgrades, he could put the new one in. New ISDN
didn't get merged until mid-2.3, and that will *not* go in 2.2.

It's a release management thing. You just *can't* break your userspace
in the middle of a stable kernel, no matter how much it seems to make
sense, unless there's a reason as important as, say, security.

> -bool 'Multiple devices driver support' CONFIG_BLK_DEV_MD
> +bool 'Unmaintained multiple devices driver support' CONFIG_BLK_DEV_MD

Looks good to me. (Not that my opinion is worth anything in this
context.)

Peter

--
Unsubscribe?  mail -s unsubscribe debian-user-request@lists.debian.org <
/dev/null

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon Feb 07 2000 - 21:00:12 EST