Before you apply a SQL Server patch, it is usually advised to perform the below:
1. Ensure that there are no user/application connections to the SQL Server instance.
2. Ensure there are no SQL agents jobs, that may kick off just before you start the patching.
3. If this is a Virtual Machine (VM), take a snapshot otherwise, ensure you at least have a backup of the system databases.
4. Restart the server. (This one step surprisingly usually resolves 99% of patch application issues and it’s not mentioned by Microsoft). Just stopping/starting the services or reviewing/resolving the locked files warning is typically less effective than a server restart.
Of course if you are using Database Mirroring, AlwaysOn, Clustering … there are other considerations, and typically these involve patching the Passive instance to minimize the outage window. Once again, restarting the passive server prior to patching usually ensures the smoothest patching experience.
Once patching is completed, it is also recommended to again restart the server. This final restart can give you the confidence that future restart issues are likely not related to this patch.
On none HA instances, these server restarts will likely increase the outage window, but it could prove to be the difference between a successful or failed patching attempt.
Previous
Next