On 07/22/10 05:56 PM, Jason wrote:
At a previous job most of their Sparc systems were upgraded from
Solaris 2.6->8->10 via live upgrade. Obviously new systems got the
latest standard, and not every system that went from 2.6->8 was still
around to do the 8->10 upgrade, but at one point we had around 1200
sparc systems (all servers, no desktops) that we maintained in our
department (with three other groups of similar size). The reduced
downtime of the update was _critical_ in allowing this to happen
(otherwise the business would force us to run ancient versions
forever). This was probably one of the few things that kept them from
tossing Sun out the door completely (which at one point they were
trying to do) -- since they were obsessed with system availability (to
a sometimes absurd extent), it provide a very distinct advantage over
the AIX and HP-UX systems they had.
I suspect they would be quite disappointed (to put it mildly) if there
is no way to do something similar (at least an installer that can run
in an older version to lay down the bits in unused space).
As Bart said, most don't do this, though I am aware of at least one
customer who does (probably the same one, based on your description).
Honestly, they're doing what we would have recommended for everyone but
didn't work hard enough to get widely accepted. I'm sure we'll work
with them to find an acceptable solution to their needs, but it's more
likely to be a special case, not a general one.
Dave
Since the hardware, OS versions (including patches), as well as other
software was tested and controlled rather carefully, we had very few
problems with this.
On Thu, Jul 22, 2010 at 4:44 PM, Ian Collins<i...@ianshome.com> wrote:
On 07/23/10 06:44 AM, Bart Smaalders wrote:
On 07/21/10 15:25, Ian Collins wrote:
If Solaris Next is to be IPS based, I really really hope we will see a
viable upgrade path. The lack of one is the biggest hurdle to IPS
adoption.
In general, upgrading from UFS root w/ svr4 packages to ZFS root w/ IPS is
a very difficult problem. While we can imagine cases in which we
could be successful, there are a host of situations which are not
readily addressable. Add to this the fact that most of Sun
^H^H^HOracle's Solaris customers generally do NOT upgrade from one
release to the next, because of the reproducibility problem - production
machine configurations need to be readily reproducible, and upgrading
an existing S10 patched OS is not the best way of doing this.
I can see that's probably true. I only ever upgraded one production box
from Solaris 9 to 10 and that was a very simple configuration.
Nearly all of the other production Solaris 9 boxes I've replaced have been
migration for their services to Solaris 10 zones. The small remainder have
been imported to branded zones. So I guess Robert is right, a branded zone
is one option!
I do wonder how much of a selling point (to keep people on Solaris) the
ability to upgrade was, even it wasn't used?
We anticipate developing and sharing migration strategies and tools,
but a traditional upgrade in place DVD approach is not likely to occur.
That's good.
The traditional upgrade in place DVD approach has probably reached the end
of the line with the increasing use of virtualisation. Treating a system
(or zone) as a service and looking at how to migrate that is better
approach.
--
Ian.
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
