From: KYLAM Tru64 (lamkaiyeung@hotmail.com)
Date: Sat Aug 02 2003 - 08:00:24 EDT
Sorry for the late summary.
Thanks Alan Rollow & Martin Petder for their prompt responses.
Alan Rollow
============
The usual choice is to upgrade the operating system
to the target version first. That ensures the rest
of the configuration is stable, so that any problems
that come up are almost certainly the result of the
upgrade. Once you have the new version stable, the
next choice depends on how much the larger, faster
system vs. the storage upgrade.
Upgrading the storage is a little easier if you have
enough switch ports to have the MA8000 and EVA 3000
at the same time. That allows you to migrate the
data from one to the other. For the data that is
on an AdvFS domain, you can even do this online
(though maybe not the root file system).
Another interesting choice, if you're going to have a
cluster anyway, is to upgrade the operating system,
turn the ES40 into the first node of the target
cluster and verify the cluster is stable. Then add
the GS1280 node(s) as members of the cluster. Once
the everything is in the cluster and it is stable,
you can remove the ES40.
The other usual choice to an upgrade like this is to
install V5.1B on the GS1280, install any 3rd party
software that needed and run the new system and old
system side by side for a while. Once you know the
new system is stable, migrate the data over and make
the new system the production system instead of the
"test" system.
Martin Petder
============
<1st upgrade to 5.1B (Can't boot marvel with previous versions)
<Then apply all relevant patches.
<Then whether migrate with AdvFS tools all disks to EVA3000 (in case
<you have the license. In case you dont - get one. It'll be highly
<usable for EVA3000 (especcially for disk resizing)). If you can't
<get a license, then copy all the data and then follow the
<root partition migration instructions available in the archives of
<this list and also (hopefully) in HP Tru64 documentation website.
<
<Finally boot Marvel with generic kernel and create a new kernel. You probably
<can't use Marvel partitioning (at least with memory striping) with migrated
<system, but at the moment it's unsupported by HP anyways.
<
<Be sure to check that all your hardware options in ES40 are also supported
<with Marvel.
<
<Also be careful while configuring EVA - it's a little different than MA,
<especcially on the LUN numbering (in case of MA the virtual disk name is the
<LUN name (apart for partitioning), but in case of EVA you have to specify
<your LUNs in advanced configuration, unless you're happy with EVA assigning
<LUNs in order of virtual disk representation creation (i.e. from 1-whatever:)
<
<Also in EVA you can't use LUN0 for data disks.
Regards
KYLAM
Original Post:
=========================
Hi Managers,
Here we have a big project.
Existing: 1 ES40 standalone, Tru64 5.0 using MA8000
Going to upgrade to: 1 GS1280, Tru64 5.1B using EVA3000
Anybody has any ideas how to upgrade the server under minimum risks and downtimes?? This server is highly critical and we can't afford any problems.
Thanks.
KYLAM
_________________________________________________________________
Linguaphone : 學英語，免費學日語 http://go.msnserver.com/HK/30475.asp
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:29 EDT