[WBEL-devel] comps.xml, warning in anaconda/pkgorder

Jon Peatfield J.S.Peatfield@damtp.cam.ac.uk
Tue, 31 Aug 2004 23:57:43 +0100


I tried to post this a few days ago but I did so from an address which
wasn't subscribed so it seems to have got stuck/lost somewhere.

Sorry if this should go to wbel-users I'm a wbel newbie and don't yet
know where the boundary between use and devel lies.

For the past few days I've been testing taking the WBEL 3.0 tree
applying the latest updates into a pxe install tree and doing a few
test installs (to check that my existing kickstart files won't need
much work if we switch to WBEL -- we are curently stuck on RH8).

All that worked fine -- I'm really impressed with WBEL, it looks like
I'll have little real pain converting our existing stuff over.

Today I decided to re-do that but starting from the respin-1 tree
(which I could only find ISOs of, is there a correct place to rsync a
tree from?).

The scripts I was using (applying the same updates as before) worked
(modulo the perl rpmvercmp I was using which thought that 15.EL was
newer than 15.0.4.EL).

However, when I came to generate the orderlist I got some new errors
from pkgorder:

  Base references package acpid which doesn't exist
  Graphical Internet references package mozilla-psm which doesn't exist

When installing from this updated tree anaconda also reports the same
errors on VC3.

No acpid rpm existed in either the 3.0 or 3.0-respin1 tree so I don't
quite understand why it is listed in the comps.xml for respin1, though
I see that there is a srpm in the respin1 tree.

So should the rpm be present or should it be removed from comps.xml?

mozilla-psm was in the old 3.0 tree but not in respin1, is this
because of a packaging change with the updates?  Again I assume it
should go from comps.xml

Not that I particularly care about either (I won't be installing
either package), but I dislike annoying messages if I can avoid them.

Note that I'm being somewhat naughty and using the pkgorder script on
RH8 (since that is all I have at hand), so I already get zillions of
warnings about things matching the following pattern:

   'Component [A-Z].*needs package.*which doesn'

which I *assume* would go away if I used a more recent version of
anaconda to do that step.

Assuming "it's just python" I did try extracting the relevant scripts
from the anaconda, anaconda-runtime rpms (WBEL 3.0 updates), but the
newer versions seem to require a bunch of extra python modules which I
don't have.  I'll try this all again once I've got a stable wbel box
to test from...

[ aside re mirrors; on 1st August the name miror.ac.uk changed hands,
the old setup (now running as mirrorservice.org) is still mirroring
whitebox (though their rsync server seems slightly ill today).  The
new mirror.ac.uk *claims* to have a mirror of whitebox but seems to
only have part of the 3.0 tree (~400M of update SRPMS).

http://www.whiteboxlinux.org/download.html lists
http://www.opensourcemirrors.org/ but that doesn't seem to be a valid
name.

I don't want to overload mirror.physics.ncsu.edu::whitebox but it *is*
the only one listed as providing rsync access.  
]

 -- Jon

Jon Peatfield,  Computer Officer,  DAMTP,  University of Cambridge
Mail:  jp107@damtp.cam.ac.uk     Web:  http://www.damtp.cam.ac.uk/