[ Saturday, January 22, 2000 ] Jakob �stergaard wrote:
> It seems improbable that the 0.90 raid stuff will go into 2.4 currently.
> This is catastrophic. We need to try Ingo's patch, and give it all
> the testing we can. If enough people try the patch and we can give a
> reasonable argument that it is stable - possibly after ironing out any
> errors there might be - it _could_ go into 2.4. We want this to happen.
Notables:
- Ingo has said in an interview that he's fine with keeping it a patch
against 2.4 and integrating in 2.5
- Alan had 0.90 RAID on the first-pass to-do list for 2.3 finish-ups
I believe that Ingo realizes that no amount of last-minute cram-testing
is going to be a real factor, it'll come down to Linus's feeling of
how well reiser, ext3, and 0.90 raid play together. If things look
(architecturally) sound, I think the amount of testing will be a secondary
issue (2.2.0 had a good amount of things shoved in at the last second).
This patch and lkml threads seem to indicate that Ingo, Alan, and Stephen
are largely in agreement on the fundamental architecture, but I think
Ingo realizes that the argument against 0.90 RAID in 2.4.0 is too easy
to make to try and fight against.
> It is important that we get a decent RAID subsystem in the 2.4 kernel.
Absolutely, but keep in mind that while we (on this list) despire
0.4x RAID, it is still running on many systems (more than we'd like :)
and when these systems want to upgrade to their newest distro version,
the distributors will have one more issue to deal with (albeit a simple
mkraid --upgrade and new raidtools). RH has taken the "plunge" and
moved their arch over to 0.90, but they seem to be one of the few and
putting things into the mainstream kernel does largely force them onto
the distributions and thusly their users.
I guess I just want to make sure that we don't get our hopes up about
2.4.0 having RAID 0.90... I hope it happens as much as anyone, but
if Ingo's willing to maintain a patch still, I think as users we
can be willing to keep using his patches :)
> will not happen unless we give Ingo's new stuff a *lot* of testing.
Now I *would* like to see as much testing of Ingo's stuff as possible,
but more because I want it tested and rock-solid, not as part of
a marketing campaign that it should be doused in penguin pee. :)
> I will test on two systems (at least), and I beg everyone out there who have a
> spare system to try the patch too. Find the bugs, report performance
> differences to linux-kernel and linux-raid, and let's get a decent RAID
> subsystem in 2.4.
I believe the architecture has the performance issues already well
understood by the decision makers, and while testing and bug fixes
are always a good thing, dozens of mails to an already heavy-traffic
lkml of "raid 0.90 iz better, muzt include it!" won't do much to
help the cause :)
Summary: test and bang on this code as much as possible (esp. the KNI
stuff if the PIII stuff happens to be included), but do it for
the sake of good testing... don't tie it to hopes of inclusion
as I don't personally want to see people get disappointed and
disillusioned with RAID testing because of any exclusion.
I also *certainly* don't want to see any animosity towards Linus
that may be generated if it's decided to be best left out. Ingo
is comfortable with that, so we should be equally comfortable.
(Yes, this email was hard to write as I'm so incredibly in agreement
with Jakob that it kills me, but I want to bring some sanity and
keep hopes reasonable in light of very realistic possibilities :)
James
--
Miscellaneous Engineer --- IBM Netfinity Performance Development
