Re: First draft list of 2.3.x "Things to fix"

[James Manning]

[ Wednesday, January 12, 2000 ] Kristofer T. Karas wrote:
> Sean Hunter wrote:
> > On Tue, Jan 11, 2000 at 12:51:06PM -0500, Kristofer T. Karas wrote:
> > > * "Got md request, not good..."
> > Could someone who knows just inform us why the "new" raid stuff that I
> > have had working on my production servers for at least 6 months isn't
> > in the dev kernel?
> 
> It is in the devo kernel, and it works fine in kernels up through 2.3.31
> (and .32?).  The "problem" is that somebody made a patch which broke MD,
> and said patch author never offered to fix their handiwork.

Wrong.  The 0.90 RAID code has yet to exist in the 2.3 tree, although
Ingo is still working on it.  You can check the linux-raid archives
for posts from him and Tweedie that address the issues very well.

Ingo recently release the new RAID code against 2.2.14 in case you want
to use that as a base http://people.redhat.com/mingo/raid-2.2.14-B1

> Where I work, there's an "if you break it, you fix it" mandate.  (And if
> you can't fix it, you come forward and offer those who can some help.)
> But that's a minority opinion on L-K, where the cutting-edge emphasis
> puts technology several steps ahead of the stable bug-fixes.  I'm not
> complaining, mind you; the SCSI mid-level code cleanups and so forth are
> much needed, and I'm glad they went in.  But I think the MD breakage has
> gone on long enough now that the authors of the new technology should
> make a token effort to spruce up the fringes of their efforts.

The issues involved are far from trivial, but the work isn't impossible
and it's getting done.  Alan has listed RAID 0.90 on his 2.3.x TODO list
and Ingo is in agreement that the new raid code "has to be merged into
2.4 or 2.5 (as soon as possible)"

Rest assured the current MD state isn't just being ignored :)

James
-- 
Miscellaneous Engineer --- IBM Netfinity Performance Development