[WBEL-users] LVM update breakage

[bishop]

Kirby C. Bohling wrote:

> On Tue, Sep 21, 2004 at 03:55:35PM -0500, Kirby C. Bohling wrote:
> 
>>On Tue, Sep 21, 2004 at 03:04:54PM +0100, Matt Dainty wrote:
>>
>>>Hi,
>>>
>>>Has anyone noticed that the updated lvm package has broken? It's now
>>>installing all of the binaries with an '.lvm1' suffix which I'm fairly
>>>sure is the reason a remote box that I updated has not come up, (It has
>>>/usr, /var, etc. on LVM).
>>>
>>>I can see this is either:
>>>
>>>1) Package breakage, they shouldn't have the '.lvm1' suffix
>>>2) The SysV script should be updated to try binaries with the suffix as
>>>well
>>>3) There's perhaps an LVM2 package that should be installed, (certainly
>>>didn't get pulled in by deps).
>>>
>>>Any comment? I'm going to now try faxing a shell script :-) to put
>>>compat symlinks back in which to my eyes will get the box working again.
>>>
>>
> 
>>WBEL, I'll have to look at the package to be sure.  I'll bet the
>>.lvm1 suffix is a "%define", that RHEL overrode on the command line
>>when they built the package.  At least I sure hope they did,
>>otherwise, I'm having a harder and harder time seeing why they put
>>the "Enterprise" part in the name.
> 
> 
> I got it wrong, it's not a "%define", it's some script.
> 
> Opps!  Didn't John say that sometimes he builds the SRPM's in a
> chrooted environment on a 2.6 kernel on a FC release.  I'd bet that
> he built the lvm packages on a non-2.4 kernel.
> 
> Here's a cut and paste out of the lvm.spec file:
> 
> # Conditionally suffix all binaries with .lvm1 and change manual
> # page names
> OS_VERSION=`uname -r`
> if [ "${OS_VERSION:0:3}" != 2.4 ]
> 
> I think he'll need to re-roll the package while running a 2.4
> kernel.  I'm building it here local, but I have to run to a dentist
> appointment before I can report the results.

Aaargh.  No way!  He should

- OS_VERSION=`uname -r`
+ OS_VERSION=%{?kernel}%{?!kernel:`uname -r`}

instead.

That sucks.

  - bish