On 3/5/2009 11:32 AM, Robert P. J. Day wrote:
> On Thu, 5 Mar 2009, David wrote:
>> On 3/5/2009 7:55 AM, Robert P. J. Day wrote:
>>> here's the current situation if anyone wants to suggest tests. i
>>> have my two gateway MX7120s installed with fedora 11 alpha -- one with
>>> i386, the other with x86_64 -- virtually identically (about 1550
>>> packages on each).
>>> the first difference is that, at the end of the install, when i
>>> rebooted, the i386 install rebooted cleanly while the x86_64 install
>>> flashed a screenful of kernel diagnostics on its way down, but it
>>> still booted cleanly coming back up again. not sure what that was all
>>> about.
>>> at this point, before i do *anything*, should i take note of the
>>> current setup in any way? suggestions? my first step would be to
>>> update the yum/rpm packages but, to do that, i would first update
>>> "fontpackages-filesystem" since its original packaging conflicted with
>>> "rpm-build" which i installed, so my proposed first step would be:
>>> # yum update fontpackages-filesystem
>>> # yum update rpm\*
>>> # yum update yum
>>> does that make sense? but even before i do that, i'm still open to
>>> what i should record in terms of current config.
>>> thoughts?
>> Do I understand correctly that these are *not* updated to current
>> Rawhide yet?
> you understand that correctly -- they are fresh off the original
> DVDs.
>> If so you would probably need to do the x86 and x86_64 updates in
>> small stages. Rawhide x86 was rebuild from i386 to i586 recently so
>> I would think that most, if not *all*, of the packages installed
>> would be an 'update' on the x86 install.
>> That would hold true for the x86_64 install too for the x86 packages
>> installed there.
> and yet, that was one of the issues -- whether (after upgrading the
> fundamental yum and rpm* packages), i should simply be able to type
> "yum update" and go drink beer. some people claimed that they had
> done exactly that -- my experience was otherwise.
> so what's the story here? are you saying that that should *not* be
> expected to succeed? if not, that's pretty important information.
Actually it should succeed but I don't think I would "go drink beer". I
might *stay* and drink beer though. 8-)
The way the update works is that *all*, you said 1500, packages would be
downloaded, then installed, and then the cleanup would begin. That would
take a lot of disk space with the potential to fail at any stage.
For the 64bit you would get *all* of the i386 now i586 packages install
and all of the 64bit packages that have changed. And Rawhide changes a
lot every day.
Someone posted a script a while back, might have been here, that was
supposed to help someone with a disk space problem by doing the update a
little at a time. IIRC it was for a release upgrade but I don't see why
it would not work here.
Disclaimer here: I have not used this. :-)
---------------------------------------------------------------------
#! /bin/bash
yum -q check-update > /root/check-update
for i in `awk '{print $1}' < /root/check-update`
do
yum -y update $i
yum clean all
done
rm /root/check-update
---------------------------------------------------------------------
And I would do this in level 3 and not in the GUI level 5. Less things
to go wrong that way.
You do know how to get to level three?
--
David
--
fedora-test-list mailing list
fedora-test-list@xxxxxxxxxx
To unsubscribe:
https://www.redhat.com/mailman/listinfo/fedora-test-list
