August 28, 2003 at 9:36 am
I inherited a SQL Server 2000 Sp2 cluster on Win2K SP2. To apply a service pack, I have learned, the server that it was initially installed to must be in control first. Also, I have learned applying SP3 and security patches will apply it to both nodes. So, to apply a service pack should I log onto the server it was installed to to apply the SP and NOT the virtual servername? Should I stop the SQL Server services via Cluster Admin (MSDTS, SQL Agent, MSSQL) before applying SP? I know SQL Server pretty well but have no experience with clusters.
August 28, 2003 at 2:58 pm
Mark,
AFAIK you should login to a node which currently owns the instance you're going to upgrade. I'm not sure if it really has to be an node where the virtual server was installed. Sp3 will update binaries on both (all) nodes that can own the instance.
I'd suggest to login to the physical node in case the virtual IP address resource goes offline during the upgrade.
You don't need to stop SQL Server. In fact you need to make sure that SQL Server is online as well as all other cluster resources in that group. There is an article on Technet about deploying SP3 on clustered SQL Server 2000: http://support.microsoft.com/default.aspx?scid=kb;en-us;811168
SP3a setup will take SQL Server offline and will restart it several times. At the end it'll reset and restart all clustered resources.
August 28, 2003 at 3:16 pm
I just did this very thing. The install does have to be on the "hot" node, but it doesn't matter whether that was the "original" node. For the cluster install the server should be online (whereas for standalone you take it offline first). Make sure any components that may get upgraded (MSDTC for example) are currently on the same node. Make sure there are no resources in the server group that might cause a failover when the sql server resources go offline, because that will screw up the install and the server goes offline several times during the install. You can just put them offline for a moment, if you need to.
Hope this helps,
Graham
August 29, 2003 at 6:05 am
OK, thanks for the help! In reviewing the document it talks about removing all depenancies from the SQL Server resources. We created a new 'drive' in which I have created a new database on it. I had to make this drive resource depenent on the MSSQLServer service. Will this be effected?
- - - - - - - - - - - - - - - - - - - - -
Verify that there are no dependencies other than those created by the SQL Server setup on any SQL Server cluster resources. While performing the cluster portion of SQL Server 2000 SP3 or SP3a, the setup will reset all cluster resources to the default properties for all resources in the group, whether they are related to SQL Server or not.
August 29, 2003 at 7:13 am
Hi. It's a mistake to make a database disk dependent on the sql server resource, for the following reasons:
1, the disk will always be brought on line AFTER the SQL Server resource, so the server will not find the database available when it starts up; and,
2, if the server fails the disk will immediately be brought off line, which will prevent you from getting access to the raw database files.
The disk shouldn't really be dependent on any resource, but if you want it to be, make it dependent on the server's network name. That way you can provide a file share for it for remote access.
August 29, 2003 at 7:16 am
p.s. The disk needs to be in the same group as the SQL server, though. That ensures it moves from node-to-node with the server.
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply