I am in the planning process of upgrading the hardware on
our cluster. I need info/advice/suggestions on how to go
about this.
I am running an active/active cluster on 2 ML370 servers.
They are running Windows 2000 Advanced Server/SP2, MSSQL
2000 Enterprise/SP2.
I want to know if all I have to do is Failover all
resources from Server2 to Server1, then evict Server2?
Then do all I have to do is run the cluster install? I am
not sure what needs to be done with the SQL instance from
Server2? Do I uninstall it or what? I hope this is clear
enough. Any ideas please let me know.

Jeff

Been there, Done that.

You will need to follow the instructions in BOL on how to remove the SQL
installation from a failed node. (Pretend one node failed and you are
removing it). This is an extra step above the eviction from the cluster
itself. Then follow the instructions again on installing SQL to a
replacement node. Be sure and read the README file on the service pack
installation to avoid causing downtime during the re-apply of a service
pack. Repeat for the other node.


--
Geoff N. Hiten
Microsoft SQL Server MVP
Senior Database Administrator
Careerbuilder.com


"Jeff" <anonymous (AT) discussions (DOT) microsoft.com> wrote

Depends on what you need to upgrade,
if it's just cpu or memory or hardware that shouldn't impact the running of sqlserver, you don't have to reinstall sql or even evict the node. you can just move all the sql instances to one node (in off-peak time, assume one node can still support 2 instances), upgrade the idle node and rejoin it when the upgrade is done.

reinstall sql or even evict the node. you can just move
all the sql instances to one node (in off-peak time,
assume one node can still support 2 instances), upgrade
the idle node and rejoin it when the upgrade is done.
replace the physical box. I am pulling server2 out and
replacing it with a newer box.

Geoff,

I dl and checked the SQLBOL.
So to make sure I got this straight:

1. failover to server1
2. run sql setup and remove server2. (remove the named
instance in setup)
3. evict node2 from cluster
4. build new server, install os, join cluster.
5. run sql setup on server1 and add server2 back to the
failover cluster?

Is this correct? Also, in doing step 5, does it copy the
sql binaries onto server2 automatically?
i am not worried about getting the server into the cluster
the only part i am confused/worried about is the sql
instance on server2. Is it a brand new install or does it
retain the information from before?
Any clarification you can provide would be appreciated.

TIA,

Jeff

You are correct. Step 5 will install the Binaries on the new node. Don't
forget to run any service pack installs directly from te new node before you
fail SQL over to it. That wil update the local binaries to the correct
service pack level.

--
Geoff N. Hiten
Microsoft SQL Server MVP
Senior Database Administrator
Careerbuilder.com

I support the Professional Association for SQL Server
www.sqlpass.org

<anonymous (AT) discussions (DOT) microsoft.com> wrote