UPDATE Re: [Cialug] upgrading opensuse 10.1 -> 10.2

Dave J. Hala Jr. dave at 58ghz.net
Thu Apr 5 16:01:17 CDT 2007


I'd state that I've never in 5 years had an update problem on any of my
Rhel servers, but that would jinx me, so I won't say it.

On Thu, 2007-04-05 at 14:45, David Champion wrote:
> Yeah, that's why RedHat servers are such a pain to update, because you 
> have to install so much extra stuff to make them useful.
> 
> Mandriva, for all it's other faults, has just about everything I need in 
>   the installs, or one of the supplemental sources (contribs or PLF). 
> It's been pretty rare that I have to install something from source.
> 
> -dc
> 
> Josh More wrote:
> >  I was also badly burned in my Red Hat / Fedora / Debian (though the
> > debian upgrade got to be OK, when they decided to stop releasing new
> > distributions (removes tongue from cheek)).
> > 
> > It seemed to me that I do so many quirky experimental things to my
> > systems, that it's unrealistic to expect the distributions to be able to
> > handle it.  What I do instead is:
> > 
> > 1) Whenever I install something on my system that is not part of the
> > "core", I make a directory inside /root/installs, and put the original
> > tarball.
> > 2) Once the item is installed, I use the 'history' command to generate
> > the list of steps that I used to install it.  This goes in
> > /root/installs/<app>/NOTES
> > 3) If I add additional repositories, (rpmforge, packman, etc), the
> > instructions go in /root/installs/repos/NOTES
> > 4) Periodically, I run 'rpm -qa | sort > /root/installs/rpms.txt'
> > 
> > Then, when it's time to upgrade, I back up /etc, /home, and /root, and
> > do a clean install.
> > That way, I have an easy reference for the strange stuff I may want to
> > do again, and a reference for what I used to have installed.  I've not
> > had an upgrade problem since I started doing things this way.
> > 
> > Now, that said, it does take some care to follow the basic rules:
> > 
> > 1) EVERYTHING on the system must be linked to an rpm (or deb, etc etc
> > etc)
> > 2) EVERY rpm (or deb, etc) installed on the system that did not come
> > from a yum (or yast, apt-get, red-carpet, etc) repo, must be in a
> > /root/installs/ directory
> > 3) EVERY tarball installed on the system must first be packaged, and
> > the packaging notes be in a /root/installs/ directory.
> >  
> > But if you can do that, most problems magically vanish.  Also, it's
> > been interesting that as the third-party repos have matured, my
> > /root/installs directories have shrunk dramatically.
> > 
> > 
> > 
> > 
> > 
> > -Josh More, RHCE, CISSP, NCLP 
> >  morej at alliancetechnologies.net 
> >  515-245-7701
> > 
> > 
> >>>> David Champion <dchampion at visionary.com> 04/05/07 2:25 PM >>> 
> > Not exactly that one, but I've been burned before by various RedHat & 
> > Mandriva system updates.
> > 
> > I've had nearly identical systems where one update goes just fine, but
> > 
> > another one fails. It's very frustrating. I've learned that any time
> > I'm 
> > going to do an update, I backup, attempt the update... if it fails,
> > then 
> > I just do a clean install and restore my backup. Probably better to get
> > 
> > a clean install anyway.
> > 
> > Most recently, after doing a Mandriva update on a system running 
> > software RAID, the update script munged the mdadm.conf and left all the
> > 
> > volumes off the DEVICE list. Needless to say, the system couldn't find
> > 
> > the RAID when it rebooted. That was a joy to track down. And it was my
> > 
> > personal server I co- lo with a really nice dude, so I had to bug him
> > to 
> > get to the box when it failed.
> > 
> > Mandriva 2006 is just about to EOL, so I'm going to have another round
> > 
> > of updates. :)
> > 
> > - dc
> > 
> > Matt Patterson wrote:
> >> Ah, the joys of system administration.
> >>
> >> I found the general source of my issues with the upgrade and grub 
> >> problem. It was one of the grub files other than menu.lst.    I'm 
> >> guessing the stage1 or 2 files.
> >>
> >> Here is how I fixed the issue.  I booted off the opensuse10.2 install
> > Cd 
> >> into rescue mode.  Mounted the root partition and brought up an IP. 
> > 
> >>  From there I went to its matching sister server and grabbed
> > everything 
> >> from /boot and /lib/modules since I was basically rolling back the 
> >> kernel at this point.  Once the files were in place on the mounted
> > root 
> >> partition, I re- ran the grub setup commands, rebooted, prayed to the
> > bit 
> >> gods and Eureka!  SUSE boots as expected.
> >>
> >> I still have a snapshot of the bad boot directory and I am going to
> > try 
> >> and see what was screwed up.  Needless to say, this is VERY annoying
> > 
> >> that this has now happened twice on different hardware.
> >>
> >> Has anyone else been bitten like this?
> >>
> >> - Matt
> >>
> >>
> >> On Thu, 5 Apr 2007, Matt Patterson wrote:
> >>
> >>> All,
> >>>
> >>> I'm starting to get that urge to kill again.  It's been one of those
> > 
> >>> nights.
> >>>
> >>> I'm having an issue with the upgrade process from opensuse 10.1 to 
> >>> opensuse 10.2. Here is what I have done.  I have a local mirror of
> > the 
> >>> opensuse repository and have set my boxes to look at the opensuse 
> >>> stable repository as described by the help section of the opensuse
> > site.
> >>> I have successfully gotten one box to go through the upgrade process
> > 
> >>> getting the patches for 10.1, reboot for the new kernel and then 
> >>> getting the rest of the packages to update to 10.2.
> >>>
> >>> Last week, on matching hardware as the 1st successful update,
> > opensuse 
> >>> failed to write out a proper menu.lst file for grub.  It put in the
> > 
> >>> initrd line but left out the root and kernel lines which, based on 
> >>> past experience, are pretty important.
> >>>
> >>> So, I got around that issue and figured it was a fluke.  That is, 
> >>> until tonight happened.  Tonight's issue is happening on different 
> >>> hardware (IBM instead of Dell).  Instead of getting a screwed up 
> >>> menu.lst file, I get a 'Grub loading, please wait' line to flash 
> >>> across my screen before the system reboots.    I booted up a rescue
> > 
> >>> CD, got into grub and did the normal root and setup command.  It 
> >>> states that everything went fine.  But I'm still in the reboot loop
> > hell.
> >>> With different hardware between the issues, I'm starting to think
> > that 
> >>> there is a bug in suse.
> >>>
> >>> Any thoughts on how to get this box back and what I can do to avoid
> > 
> >>> this issue in the future?  I'm going to try lilo in the morning and
> > 
> >>> see if that will allow this thing to boot properly.
> >>>
> >>>
> >>> - Matt
> >>> _______________________________________________
> >>> Cialug mailing list
> >>> Cialug at cialug.org
> >>> http://cialug.org/mailman/listinfo/cialug
> >>>
> >>>
> >> _______________________________________________
> >> Cialug mailing list
> >> Cialug at cialug.org
> >> http://cialug.org/mailman/listinfo/cialug
> >>
> > 
> > 
> > _______________________________________________
> > Cialug mailing list
> > Cialug at cialug.org
> > http://cialug.org/mailman/listinfo/cialug
> > 
> > _______________________________________________
> > Cialug mailing list
> > Cialug at cialug.org
> > http://cialug.org/mailman/listinfo/cialug
> > 
> 
> 
> _______________________________________________
> Cialug mailing list
> Cialug at cialug.org
> http://cialug.org/mailman/listinfo/cialug
-- 



More information about the Cialug mailing list