[linux-lvm] Is LVM dead ?

[linux-lvm] Is LVM dead ?

[Michael Tokarev]

I see very strange situation with lvm now.
And it is really strange for me.

Ok, I signed petition, there are far too many efforts to include
lvm into kernel.  And it was included.  And that was the last news
item at lvm website!  There are too many bugs in lvmtools, many of
them already found/reported.  There are far too many headaches to
build lvm with new kernels.  There are too many patches floating
around.  And the latest file date is about march-2000...  It seemed
to be that folks has have only one target -- to include lvm into
kernel.  To set up a dots under i's.  And that was a final plan.
Again, when IBM comes with their ideas (not to say here good or
bad -- it is a different story), people here just said something
like "we already have all that we want, so go out"... (discussion
was long and interesting, but primary authors was not involved
there).

What's happening?  I'm really worried, as this is a very
useful thing.  And I call here -- should someone become a
new maintainer of the whole thing ? :(((  As current
maintainer as it seemed to be just did all that he want
already...

I'm sorry, I'm really very sorry for this all if I'm wrong.
This is just my impression, nothing more.  And I apologize,
really, if I'm wrong.

But -- who can demonstrate that I'm wrong ??? :(((((((((

Regards,
 Michael.

Re: [linux-lvm] Is LVM dead ?

[Heinz Mauelshagen]

Michael,

as mentioned on this list a couple of days ago, LVM 0.9 is going to be
released early november.
It will for eg. contain persistent and resizable snapshots.

If you have any feature enhancements, bug fixes or code to supply,
Please feel free to discuss them on this list.

On Sat, Oct 21, 2000 at 01:08:44AM +0400, Michael Tokarev wrote:
> I see very strange situation with lvm now.
> And it is really strange for me.
> 
> Ok, I signed petition, there are far too many efforts to include
> lvm into kernel.  And it was included.  And that was the last news
> item at lvm website!  There are too many bugs in lvmtools, many of
> them already found/reported.  There are far too many headaches to
> build lvm with new kernels.  There are too many patches floating
> around.  And the latest file date is about march-2000...  It seemed
> to be that folks has have only one target -- to include lvm into
> kernel.  To set up a dots under i's.  And that was a final plan.
> Again, when IBM comes with their ideas (not to say here good or
> bad -- it is a different story), people here just said something
> like "we already have all that we want, so go out"... (discussion
> was long and interesting, but primary authors was not involved
> there).
> 
> What's happening? 

Code is beeing developed, features are beeing implemented etc.

> I'm really worried, as this is a very
> useful thing.  And I call here -- should someone become a
> new maintainer of the whole thing ? :(((  As current
> maintainer as it seemed to be just did all that he want
> already...

Did you actually read the documentaion (TODO)?

> 
> I'm sorry, I'm really very sorry for this all if I'm wrong.
> This is just my impression, nothing more.  And I apologize,
> really, if I'm wrong.

You should.

> 
> But -- who can demonstrate that I'm wrong ??? :(((((((((

Hopefully i was.

> 
> Regards,
>  Michael.

-- 

Regards,
Heinz      -- The LVM guy --

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Bartningstr. 12
                                                  64289 Darmstadt
                                                  Germany
Mauelshagen@Sistina.com                           +49 6151 7103 86
                                                       FAX 7103 96
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] Is LVM dead ?

[Andi Kleen]

On Fri, Oct 20, 2000 at 10:03:00PM -0500, Heinz Mauelshagen wrote:
> > 
> > What's happening? 
> 
> Code is beeing developed, features are beeing implemented etc.

Perhaps it would be useful if you could do an interim 0.8 mainteance
release, with e.g. all the simple bugfixes Andreas Dilger has collected
(segfaults etc.) There are unfortunately a few annoying bugs in the stock 
0.8 userland, which people rediscover and rereport all the time.

-Andi

Re: [linux-lvm] Is LVM dead ?

[Paul Jakma]

On Sat, 21 Oct 2000, Andi Kleen wrote:

> Perhaps it would be useful if you could do an interim 0.8
> mainteance release, with e.g. all the simple bugfixes Andreas
> Dilger has collected (segfaults etc.).

i second this. possibly Heinz as the main developer is currently very
deep into the cool new stuff of his upcoming v0.9, and so working on
0.8 isn't very attractive to him.

however, for users, 0.8 is what we have and it's become very difficult
to keep track of all the different patches/fixes. (least it would be
if it wasn't for andreas).

> There are unfortunately a few annoying bugs in the stock 0.8
> userland, which people rediscover and rereport all the time.
> 

also, there are people who'd love to try LVM out on bigger machines
than home servers and laptops. but who need to convince a boss
first. a clean install process makes that latter part a lot
easier. 

> -Andi
> 

--paulj

Re: [linux-lvm] Is LVM dead ?

[Jan Niehusmann]

On Sat, Oct 21, 2000 at 05:49:24AM +0200, Andi Kleen wrote:
> Perhaps it would be useful if you could do an interim 0.8 mainteance
> release, with e.g. all the simple bugfixes Andreas Dilger has collected
> (segfaults etc.) There are unfortunately a few annoying bugs in the stock 
> 0.8 userland, which people rediscover and rereport all the time.

Actually I think Andreas could do the interim releases (if he wants to).
There is nothing wrong with having one maintainer for the development tree
and one for the stable tree. Andreas already has a .rpm with all the known
bugfixes, all we need now, is a tarball of that stuff.

And the 0.8 userlevel tools will be important: The 0.9 kernel changes are 
probably fairly huge, so Linus may not include them in 2.4. That means we
need stable 0.8 tools for 2.4.

Jan

Re: [linux-lvm] Is LVM dead ?

[Michael Ju. Tokarev]

Heinz Mauelshagen wrote:
> 
> Michael,
> 
> as mentioned on this list a couple of days ago, LVM 0.9 is going to be
> released early november.

I follow this list for about a year now.  There was some references
last month to 0.9, two of them was yours.  But that 0.9 is like a
cloud that maybe here or maybe not.  In contrast, demands for
bugfixes are high for some months already.  Most important for
many users are bugfixes -- with current features lvm already
very useful, but probably not so useful with bugs and great features.
I say here about already well-known bugs that are serious enouth.
Can someone count how many people debugged existing bugs that
was already fixed in some 0.9 code? :(  Look to list: there is
a loop like:

  can someone point me to lvm for kernel 2.2.15 (2.2.16, 2.2.17,...)?
    -- yes, it is there and there
       -- thanks
  oh, there is a bug in lvm there and there.  Here is a patch.
    -- yes, it is a known bug, patch already there and there and there.
  oh, there is another bug, patch included.
    -- yes, it is also a known bug, patch can be found...

And so on.  Most people just stuck on applying kernel patches, as
current kernel is different compared to 2.2.13 one...

Great thanks to Andeas Dilger who maintains his version of all
needed stuff -- I'm shure that this work saved a lot of the above
cycles.

> It will for eg. contain persistent and resizable snapshots.
> 
> If you have any feature enhancements, bug fixes or code to supply,
> Please feel free to discuss them on this list.

What really needed is a collection of all patches or, preferrably,
one (maybe inetrim) release, with a point to it from
linux.msede.com/lvm.

[my flaws skipped]
> > I'm sorry, I'm really very sorry for this all if I'm wrong.
> > This is just my impression, nothing more.  And I apologize,
> > really, if I'm wrong.
> 
> You should.
>
> > But -- who can demonstrate that I'm wrong ??? :(((((((((
> 
> Hopefully i was.

Ok.  I greatly apologize, Heinz!  Excuse me please.

But you should agree that from "outside" view things
_may_ looks like I described.  I has some talks with
my friends today, and some of them have the same
impressions...  And look -- situation is trivial
to reslove, and many people will be glad to help.

Also I'm shure that someone will be glad/interested
to help with 0.9 release implementation...

Again, Excuse me please, Heinz and all!

Regards,
 Michael.

Re: [linux-lvm] Is LVM dead ?

[Andreas Dilger]

Jan Niehusmann writes:
> On Sat, Oct 21, 2000 at 05:49:24AM +0200, Andi Kleen wrote:
> > Perhaps it would be useful if you could do an interim 0.8 mainteance
> > release, with e.g. all the simple bugfixes Andreas Dilger has collected
> > (segfaults etc.) There are unfortunately a few annoying bugs in the stock 
> > 0.8 userland, which people rediscover and rereport all the time.
> 
> Actually I think Andreas could do the interim releases (if he wants to).
> There is nothing wrong with having one maintainer for the development tree
> and one for the stable tree. Andreas already has a .rpm with all the known
> bugfixes, all we need now, is a tarball of that stuff.

If you don't have the rpm tool (which can also be installed on a Debian
system), you could also use alien to get the .tgz out of the .rpm file.
If Heinz agrees to link my archives from the LVM site, then I will even
make a separate patched .tgz available along with the .rpm.

> And the 0.8 userlevel tools will be important: The 0.9 kernel changes are 
> probably fairly huge, so Linus may not include them in 2.4. That means we
> need stable 0.8 tools for 2.4.

The one thing that needs to happen for my userland tools to work is to
patch the 2.4 kernel lvm.h to allow being used from kernel and user space.
I doubt that Linus would accept a patch from me on this, so it would need
to go through Heinz.  The total number of real changes is very small compared
to the 2.4 kernel header, but there are a lot of variable types that change
to be compatible with both user and kernel includes, and a bunch of format
changes.

Cheers, Andreas
-- 
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert

Re: [linux-lvm] Is LVM dead ?

[Andreas Dilger]

Michael writes:
> There are too many patches floating around.  And the latest file date
> is about march-2000...  It seemed to be that folks has have only one
> target -- to include lvm into kernel.  To set up a dots under i's.

I agree that there should be more visible interim releases of LVM
user tools.  The kernel code is now part of 2.4, so it doesn't make
sense to make 2.4 kernel patches available anywhere else.  It may be
that we can get LVM 0.8 into the 2.2 kernel - I don't know if anyone
has tried yet.  In any case, I will continue to keep the 2.2 code
up to date.

> Again, when IBM comes with their ideas (not to say here good or
> bad -- it is a different story), people here just said something
> like "we already have all that we want, so go out"... (discussion
> was long and interesting, but primary authors was not involved
> there).
> 
> What's happening?  I'm really worried, as this is a very
> useful thing.  And I call here -- should someone become a
> new maintainer of the whole thing ? :(((  As current
> maintainer as it seemed to be just did all that he want
> already...

Actually, just to let you know what is happening, a lot of people
interested in LVM got together in Miami for the Storage Workshop
(Heinz was one of the organizers there), and talked about future
LVM development.  IBM was there, as well as a number of kernel
developers, Heinz, myself (for TurboLinux), LinuxCare, etc.  It
may not be visible that things are going on, but they are.  I'm
sure there will be more information available here when it is
more than just talk.

Cheers, Andreas
-- 
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert