We have a 2-server clustered (Active/Passive) SQL 2008 R2 RTM environment we need to update to SP2 along with Cumulative Update package 1. While updating SQL Server, it is essential we minimize the downtime to our users. For planning, I am trying to determine when and for how long SQL services will be offline.
I have already reviewed the MS KB article describing the failover cluster rolling patch/SP process (http://support.microsoft.com/kb/958734
), but that has left me with questions.
This article, as I have read it, says to apply the updates/patches to the nodes that are no longer possible owners of the clustered resources. In which case, it would not longer have access to the Quorum or the shared disk storage where all database files are located (master, msdb, model, and all user db's).
Referring to the system databases, MS also states that "Installing a SQL Server 2008 R2 update changes these databases, making them incompatible with earlier versions of SQL Server 2008 R2". (ref: http://msdn.microsoft.com/library/dd638062(SQL.105).aspx
Based on the steps in the cluster instructions (KB958734 that I referred to earlier):
- Step 2 is removing passive node(s) from the possible owners of clustered services.
- Step 3 has the updates occurring on those nodes no longer possible owners of the services.
- Step 4 is adding back the now-updated nodes to the cluster services owners list.
- Step 5 is moving the cluster services to a node that was updated.
1. At this point (Step 5) when the SQL Services come online on the updated node, is this when it writes updates to the system databases? If so, is this when there will be a delay in all services coming online while the appropriate data updates are performed?
2. Will the number and/or size of the user databases on the server impact the amount of time the upgrade will complete and services will come back online?
FYI Each cluster node is running Windows Server 2008 R2 Datacenter 64-bit, SP1 with 32 GB memory.