Plague and other distros...

[Chris Weyl]

Hey all --

Just a quick update, for those that are curious :)

RHEL -- This builds w/o issue under plague, mock configs are trivial. 
Two things:  you'll need to create your own yum repso; in rhel 3,
runuser needs to be set to '/bin/su'.

Mandrake -- Again, yum repos need to be built (if anyone knows of any
public repos with the base os and updates, I'd be much obliged). 
Ditch the "delta" rpms when generating, they'll just mess things up
(for our purposes).  I've been working with mandriva 2006.0. 
'runuser' also needs to be set to '/bin/su' here.

For whatever reason (that I haven't been fully able to deduce), I've
found that the "checking for installed but unpackaged files" bit fails
horribly under mdk.  For mdk spec files (and problematic rhel ones),
replacing that macro with this will help:

%define __check_files  /bin/sh -c "(export TMPDIR=%{_tmppath} ; cd
%{_builddir}; /usr/lib/rpm/check-files %{buildroot})"

CVS integration works nicely as well -- make plague, make tag, etc,
all work nicely if you tweak CVSROOT/branches to include the rhel/mdk
distros.  There are some functions that are fedora-hardcoded in
Makefile.common (e.g. "make mock"), but the basics work w/o tinkering.

One issue I am noting is that at the top/bottom of build.log's from
rhel, I often find 3-7 lines of:

error: Macro % has illegal name (%define)
error: Macro % has illegal name (%define)
etc, etc

I'm not entirely sure what's causing this.  The build looks to
complete successfully (with resulting rpms to show for it), but
plague/mock claims it failed, so this is a bit of a problem.

Anyways...  for the curious :)  Basic upshot is that I haven't found
any changes that need to be made to plague proper, and 99% of it is
finding/generating yum repositories and tweaking mock config files.

                                                 -Chris

[Chris Weyl]

On 11/22/05, Chris Weyl <chris.weyl(a)gmail.com&gt; wrote:
>  I'm not entirely sure what's causing this.  The build looks
> to
>  complete successfully (with resulting rpms to show for it), but
>  plague/mock claims it failed, so this is a bit of a problem. 
Hrm, scratch that.  It's job.log claiming it failed; so that's plague
thinking it failed, not mock (right?).

                                                 -Chris

[Dan Williams]

On Tue, 2005-11-22 at 16:16 -0500, Chris Weyl wrote:
>  On 11/22/05, Chris Weyl <chris.weyl(a)gmail.com&gt; wrote:
>  > I'm not entirely sure what's causing this.  The build looks to
>  > complete successfully (with resulting rpms to show for it), but
>  > plague/mock claims it failed, so this is a bit of a problem.
>  
>  Hrm, scratch that.  It's job.log claiming it failed; so that's plague
>  thinking it failed, not mock (right?). 
It depends on what plague-builder got for the return code from mock.
Plague-builder will also kill mock if mock doesn't write out its status
file after 20s.  You could put some stuff in plague-builder in the
_watch_mock() function to see what mock returns.  But if job.log claims
that something failed, then yes that's plague-builder and not mock.

Dan