Anaconda early startup failure - traceback

[whitivery]

whitivery <co55-sy1t at dea.spamcon.org> wrote:

>whitivery <co55-sy1t at dea.spamcon.org> wrote:
>
>>James Cammarata <jimi at sngx.net> wrote:
>>
>>>
>>>On Thu, 02 Jul 2009 10:54:34 -0400, Kyle Powell <kpowell at redhat.com> wrote:
>>>> -----BEGIN PGP SIGNED MESSAGE-----
>>>> Hash: SHA1
>>>> 
>>>> whitivery wrote:
>>>>> We have a working Kickstart server for CentOS 4.4, building
>>>>> targets running CentOS 4.4.
>>>>> 
>>>>> I built a new server with CentOS 5.3, to build CentOS 5.3
>>>>> targets, adapting for the changes I noticed as I went along.  The
>>>>> new server boots and runs but won't build a target unit.
>>>>> 
>>>>> When I PXE-boot a target unit, I get the boot menu and pick an
>>>>> entry, it starts the early part fine, loads stage2.img OK, shows
>>>>> a couple screens then a "Retrieving installation information"
>>>>> dialog/progress bar, and just when/after that gets to 100%, it
>>>>> halts with a traceback.
>>>> 
>>>> Looks like you need more RAM:
>>>> http://www.centos.org/modules/newbb/viewtopic.php?topic_id=19367&forum=37
>>>> 
>>>> Same thing with Fedora:
>>>> http://bugzilla.redhat.com/show_bug.cgi?id=499585
>>>> 
>>>> Minimum RAM requirement for RHEL 5 is 512MB. I assume that applies to
>>>> CentOS as
>>>> well:
>>>> http://www.redhat.com/rhel/compare/
>>>
>>>Yes I've seen this same error on systems with insufficient ram (usually
>>>VM's).  The key is that it fails on the lspci call, which is exactly where
>>>all of mine would fail.
>>
>>Thanks to you and Kyle for the answers - I tried another box with
>>512M RAM and it worked fine.  Seems nuts that hundreds of
>>millions of bytes of memory can't even start a text mode install.
>>
>>With the box that worked, I still see some warnings in the
>>anaconda.log like the following - do I need some additional
>>kickstart settings or something to avoid these?:
>>
>>  I don't need DNS hostname lookup so why is it trying?:
>>21:00:01 ERROR   : no DNS servers, can't look up hostname
>>21:00:01 ERROR   : no DNS servers, can't look up hostname
>>
>>  I specified "url" so why these messages?:
>>21:00:01 ERROR   : got to setupCdrom without a CD device
>>21:00:22 WARNING : no floppy devices found but we'll try fd0 anyway
>>
>>  I specified "text" so why does it say "not available" - it's
>>  available, I just prefer text install:
>>21:00:25 WARNING : Graphical installation not available...  Starting text
>>mode.
>>
>>  What does this mean?:
>>21:00:28 WARNING : Unable to find temp path, going to use ramfs path
>>
>>  Why these?:
>>21:00:32 WARNING : step installtype does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>21:00:32 WARNING : step complete does not exist
>>
>>  These seem to be bugs:
>>21:00:36 WARNING : /usr/lib/python2.4/site-packages/snack.py:250:
>>DeprecationWarning: integer argument expected, got float
>>  self.w = _snack.scale(width, total)
>>
>>21:00:36 WARNING : /usr/lib/python2.4/site-packages/snack.py:247:
>>DeprecationWarning: integer argument expected, got float
>>  self.w.scaleSet(amount)
>>
>>21:00:48 WARNING : /usr/lib/python2.4/site-packages/yum/repos.py:68:
>>RuntimeWarning: YumProgress.progressbar called when popped
>>  self.callback.progressbar(num, len(repos), repo.id)
>>
>>  In addition, there are numerous "WARNING: <something> doesn't
>>  exist" scattered in the log near the end of the install.
>
>Following up to my post: another problem is that %end still
>doesn't work - using it to end the packages, pre, or post
>sections, it causes a fatal failure, kickstart stops.
>
>I expected this with my old CentOS 4.4 kickstart/anaconda, but
>not with the much newer CentOS 5.3 one - the web documentation
>implies that it should work.

Yet another follow-up problem/question: in the old CentOS 4.4 the
--erroronfail didn't work.  I thought it was supposed to work in
this version (CentOS 5.3), but trying it in both the pre and post
sections, it doesn't seem to do anything.  If a command returns
non-zero, or a bad (non-existent) command is run in either
section, in the latter case there's a complaint in tty3 but in
either case no effect on the install, it keeps going.

Maybe this is how they are supposed to work, the documentation
doesn't really say exactly what happens or what is considered a
failure.