Microsoft Lync Server 2013 Front End Maintenance

The first cumulative update for Microsoft Lync Server 2013 was released in February 2013. With the release of the first update to Lync Server 2013, there are some items I wanted to cover so that you aren't taken by surprise when performing maintenance or patching of the Lync 2013 Front End servers. I'll cover the following areas related to Lync Server 2013 Front End maintenance:

Pool Quorum

Lync Server 2013 introduced the concept of a quorum, which is a major change from previous editions of Lync. Microsoft Office Communications Server (OCS) 2007 R2 introduced SQL Server clustering with Back End servers for the Front End pool. A basic understanding of how quorums work with clustering is now coming into play for Lync 2013 administrators, not necessarily on the SQL Server Back End server side but on the Front End servers. In order for a Lync 2013 Front End pool to be considered in a functional state (which is a good thing), a certain number of Front End servers in the pool need to be up and running. Table 1 shows the minimum number of servers that are required to be online and available in order for the Lync 2013 pool to be in a functioning state.

In the event that you're performing maintenance or patching Lync 2013 Front End servers and you bring more than the appropriate number of Front End servers down, you'll get an error message in the Event viewer stating "Local Pool Manager has been disconnected from Pool Manager." If the number of remaining running servers is still below the threshold level, the remaining servers Lync services in the pool will stop, making the pool inoperable.

Upgrade Domains

Front End servers in an Enterprise Edition pool are organized into upgrade domains. Upgrade domains are created automatically by Topology Builder when a Front End server is added to the topology and published. The Microsoft Lync Server team recommends when upgrading Front End servers that you perform the upgrades one server at a time, as opposed to upgrading several Front End servers at the same time. The recommended approach calls for bringing a single server down, upgrading it, and then restarting it before you upgrade another server.

From the Lync Server Management Shell, running the cmdlet Get-CsPoolUpgradeReadiness displays results if the current pool is ready to be patched or serviced. Figure 1 shows the number of Front End servers that are listed in the pool and the number that are currently active.

Figure 1: Currently Active Front End Servers

Since both servers in the pool are running, the value of True is given to IsReadyForUpgrade. From this point the administrator can proceed with taking one of the servers offline for maintenance or patching without affecting the pool's state.

The results could differ if one of the two Lync Front End servers wasn't available while running the cmdlet Get-CsPoolUpgradeReadiness from the Lync Server Management Shell. Figure 2 shows a pool that isn't ready to allow maintenance to take place.

Figure 2: Pool of Servers Not Ready for Maintenance

The value of IsReadyForUpgrade is False, which displays to the administrator. In addition, the False value also lets the administrator know that he shouldn't proceed with taking the only remaining server offline for maintenance; doing so could put the pool in an unresponsive state for users.

Best Practices

When you plan an upgrade or maintenance of a Lync 2013 pool, Microsoft's recommendation is to update one server at a time. Bring one server down, apply the upgrade, and then bring that server back up before upgrading another server. For detailed instructions on upgrading a Front End server with the Lync 2013 Cumulative Update 1, see "Updates for Lync Server 2013."

Final Thoughts

The available Lync Server Management Shell cmdlets for checking pool states provide a structured approach to performing patches and maintenance in a Lync 2013 Front End environment, despite the nuances of Front End server clustering and the need to have a certain number of servers in the pool up and running in order for the cluster to be in such a state that it's considered to be functioning. Although patching seems like a complex operation, following the proper guidance helps make the process a smooth one.