[prev in list] [next in list] [prev in thread] [next in thread] 

List:       opensolaris-install-discuss
Subject:    Re: [install-discuss] nv b95: pkginstall coredumps on nfs:/...
From:       Jens Elkner <jel+install () cs ! uni-magdeburg ! de>
Date:       2008-08-23 4:29:32
Message-ID: 20080823042932.GA25695 () papst ! cs ! uni-magdeburg ! de
[Download RAW message or body]

On Fri, Aug 22, 2008 at 07:42:53AM -0400, James Carlson wrote:
> Jens Elkner writes:
> > 
> > Must admit, that I haven't yet the time, to study live upgrade (and also
> > reading about some bugs tells me to give it not a high priority).
> 
> Many of the bugs (ironically) are in the "big servers" sorts of
> configurations.  I've been using LU on all of my desktop machines and
> my laptop (and a few lab machines as well) since the Solaris 8 days.

OK - I'll try it next time. For our sparc servers it is probably a little
bit tricky, since they have usually 2x HDD UFS SVM for the basics (i.e.
256 MB /, /var , /usr , /opt, /zones) and 2x HDD ZFS for "the real" data 
- so not really something free to do LU ...

> I've always run the latest build (upgrading every two weeks), so when
> there are bumps, I get to experience them, but I couldn't imagine
> going back to the uncertainties of standard upgrade.

OK - it is on the todo list now ;-).  BTW: The last time I installed a 
Linux distro (my home desktop + some not so big servers) dates back to 1998. 
Since than on demand I used removepkg once, and after that pkgadd/pkgrm to
keep them more or less uptodate without any problems as well ... (of course,
you can't call it slackware anymore) ;-)

> That looks like a new one to me.  It's a crash in libinst and (at a
> guess) it's a bad vfstab entry.
> > 
> > Excellent guess!
> 
> Thanks.  Wasn't hard when looking at the source.  That
> "already_mounted" function doesn't do much else.
  
Hmm - since the entries are correct, I can't see ad hoc, where the
problem should be. But I guess my sources are too old (2.8) or already
fixed, when ported to Linux ;-)

> > Since Sun sales people always don't forget to tell stories wrt.
> > backward compat. of solaris, I thought, stressing this a little bit
> > can't hurt ;-)

> The system components, though, are far more complex.  They generally
> depend on undocumented (not "secret," but just not formally documented
> or stable) interfaces, and there's no expectation at all that they'll
> work on another system.

Yepp - understood.
  
> > Thanx a lot,
> 
> No problem; glad you found a work-around.

Yepp - was easier than estimated: introduced a check in the finish
script and because all add. stuff is nfs:// a grep in the SI_PROFILE ...
~ 20 lines do the hack.

Thanx again,
jel.
-- 
Otto-von-Guericke University     http://www.cs.uni-magdeburg.de/
Department of Computer Science   Geb. 29 R 027, Universitaetsplatz 2
39106 Magdeburg, Germany         Tel: +49 391 67 12768
_______________________________________________
install-discuss mailing list
install-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/install-discuss
[prev in list] [next in list] [prev in thread] [next in thread]