(this article is under development)
It is important to test the update process in a test environment. Test environment must be as much as possible similar to production.
Applying SharePoint updates to a server farm
Safest update method
The safest method to update a SharePoint farm is to take the entire farm offline, update all servers, and then bring the farm back online. This method requires a maintenance window that might not be practical for all organizations.
High availability updating
High availability updating involves more planning, testing, and coordination. The general outline for the process includes the following steps.
(this article is under development)
It is important to test the update process in a test environment. Test environment must be as much as possible similar to production.
Applying SharePoint 2013 updates to a server farm
Safest update method
The safest method to update a SharePoint farm is to take the entire farm offline, update all servers, and then bring the farm back online. This method requires a maintenance window that might not be practical for all organizations.
High availability updating
High availability updating involves more planning, testing, and coordination. The general outline for the process includes the following steps.
- Update binaries – install patch on each server
- Update servers by role, in the following order:
- Application servers. Start with the services that are most important to the organization. (If you have no idea – update search servers before other application servers. Within the search service application roles, servers that host the index are updated after servers that host other search roles
- Web servers. Note: For each server, when the patch is applied IIS will stop and won’t resume again until the patch is installed & the configuration wizard (psconfig) is run afterwards (???).
- Database servers (content databases before search databases)
- For each role, update half the servers at a time
- Remove the first half of the servers from the rotation of the load balancer. Update these servers.
- Remove the second half of the servers of the same role from the load balancing rotation
- Immediately return the first half of the servers to the load-balancing rotation
- Update the second half of the servers and then return these to the load-balancing rotation
- Upgrade Databases. Run the Psconfig command-line tool (Configuration Wizard) on one of the servers to update SharePoint databases. This isn’t always the case – it depends on what version you’re upgrading to and from.
Enshure all servers in the farm are patched before this step. The wizard/psconfig won’t run on a patched server after patch install until all servers in the farm.
Once all servers are patched run the configuration wizard on the 1st server you want to bring back online and following instructions (or run psconfig for advanced users) - Complete upgrade – when the databases are patched, all remaining servers need to complete the patch process by running psconfig on each after the farm is upgraded.
Test environment
The advantages of building an exact-production-copy test environment include the following:
- The environment more closely resembles the production environment with combination of virtual roles shared on physical hardware.
- The performance of updating the servers will more closely match the performance in production. This will give you an idea of how long it will take to update the production environment.
- You can use procedures and processes that you develop in the test environment when you update the production environment.
Smaller environments can be used to test the update process. In any test environment, be sure to include at least two instances of every role represented on the farm.
Batch processes on a batch-processing servers
Consider to stop batch processes consumes a lot of processor time like crawling or synchronizing while updating.
Monitoring servers
Be sure to coordinate with your operations team when servers are taken offline for maintenance. Remote actions that bring a server back online during the patching process can produce unrecoverable error states.
Operating system updates
The process that updates the operating system for servers in a SharePoint farm is uncomplicated and can be performed server by server (one server at a time):
- Remove a server from the rotation of the load balancer.
- Update the server.
- Return the server to the rotation of the server farm.
This guidance applies to both physical and virtual servers.
SQL Server updates
See SQL Server documentation for guidance about how to patch SQL Server, including guidance for SQL Server 2012 AlwaysOn.
Office Web Apps Farm and Workflow Manager farm
You can update Office Web Apps Farm and Workflow Manager farm independent of updates of the SharePoint environment.
Top mistakes to avoid when you update farms
- Monitoring servers during the update process — An operations team can initiate remote actions that interfere with servers that are offline for updating. Coordinate with the operations team to remove servers from monitoring.
- Updating one server at a time — If you return updated servers to a farm one-by-one, you can cause server roles of two different versions to run in the production farm. This can compromise performance or cause error states. Instead, for each server role, take half of the servers offline and update these servers. Before you return these servers to the rotation, remove the second half of the servers that run the same role from the rotation.
- Waiting to update databases — Databases do not always need to be updated at the same time as the rest of the farm. This allows critical updates to be applied to a production farm with minimal downtime. However, databases should be updated within the next week or month.
- Not testing the update process and code
=====================================
-
SharePoint 2013 Updates
- SharePoint Patching
- http://blogs.msdn.com/b/sambetts/archive/2013/08/22/sharepoint-farm-patching-explained.aspx
- Office Web Apps (WAC, OWAS) Updates
- http://www.wictorwilen.se/WACVersions