Monitoring Microsoft Lync Server 2010 with Monitoring Studio

Downloads

Objective

Our goal is to monitor Microsoft Lync Server 2010 in BMC PATROL (and/or BMC TrueSight Operations Management) to detect failures, errors and performance problems.

This article only applies to the version 2010 of Microsoft Lync Server, but can certainly be adapted to earlier versions (like Microsoft OCS 2007 notably).

Solution

• Monitoring Studio KM for PATROL
• Monitored components:
  • 1/Windows services
  • 2/Processes
  • 3/Windows Event Logs
  • 4/Performance Counters

Installation

• Prerequisites
• Procedure
• Alternate Installation Procedure
• Post-installation tasks

Solution

Monitoring Studio KM for PATROL

Our solution relies on Monitoring Studio KM for PATROL, which is a configurable module for BMC PATROL (and therefore BMC TrueSight Operations Management). Typically, a PATROL administrator will use Monitoring Studio KM’s GUI to “create” the monitoring of anything which the administrator does not have a built-in KM for. The “setup” of the monitoring is stored in the PATROL Agent’s configuration.

More information about Monitoring Studio KM is available here on Sentry’s Website.

Here, Sentry has already built a configuration for Monitoring Studio KM that “makes” it monitor Microsoft Lync Server 2010. This article simply explains how to import this pre-built configuration and customize it to monitor your Microsoft Lync environment.

Monitored components

The architecture of Microsoft Lync Server 2010 can be quite complex as it involves several systems with different roles, where each role runs different components of Lync Server: Archive Servers, Edge Servers, Quality Monitoring Servers, Microsoft SQL Server, etc.

Note: The solution described here does not cover Microsoft SQL Server or other non-Lync components that may participate in the architecture, like NLB, MSCS, etc. The user is invited to configure other KMs specific to these technologies, notably: BMC ProactiveNet Performance Management for Servers (for Microsoft Windows infrastructure, including Windows itself, MSCS, NLB, MSMQ, etc.) and BMC ProactiveNet Performance Management for Databases (for Microsoft SQL Server)

The solution monitors a mixture of Windows services, processes, Windows Event Logs and performance counters that are grouped based upon which component of Lync Server they relate to.

1/Windows services

The status of the services listed in the table below is monitored by the solution. Each service is represented with a separate object in the console. The “Status” parameter reports the status of the service. This parameter will trigger an alarm if the service is stopped and a warning if the service is in an intermediate state (stop pending, paused, etc.).

2/Processes

Key performance metrics are constantly monitored for all of the Lync Server processes. Each process is monitored independently and displayed as a separate object in the console. The sum of all of these processes is available as a separate object as well so administrators can check the system resource consumption of Lync Server in general, and of each process individually.

Monitored processes:

• Process: ASMCUSvc.exe
• Process: AVMCUSvc.exe
• Process: DataMCUSvc.exe
• Process: FileTransferAgent.exe
• Process: IMMCUSvc.exe
• Process: MasterReplicatorAgent.exe
• Process: MeetingMCUSvc.exe
• Process: OcsAppServerHost.exe RTCATS
• Process: QoEAgent.exe
• Process: ReplicaReplicatorAgent.exe
• Process: RTCSrv.exe
• Process: w3wp.exe
• Total Lync Server Processes (all processes that have %{PATH} in their path)

For each of these processes, the following parameters are monitored:

• HandleCount
• PageFaultsPerSec
• PagefileBytes
• PrivateBytes
• ProcessorTime
• ThreadCount
• VirtualBytes
• WorkingSet

Example of process monitoring:

By default, only the ProcessorTime parameter triggers a warning when it reaches 90% five times in a row and 99% twice in a row. These thresholds (like any alert threshold in Monitoring Studio) can be customized through the GUI:

3/Windows Event Logs

When Microsoft Lync Server encounters a problem, it reports it as an event in the Windows Event Log dedicated to Microsoft Lync. These events are constantly monitored by the solution and any “warning” or “error” event related to Lync in the Windows Event Log will be reported in the console.

Each Event Log Source is represented with a separate icon in the console. For each instance, the MatchingEventCount parameter reports the number of warning and error events since the last reset of the counter. For each new event, the MatchingEventCount parameter is increased by one. As the alarm threshold is set to 1 on this parameter, an alert is triggered in the PATROL Console as soon as something wrong happens with Microsoft Lync. The MatchingEventCount keeps the same value until it is manually acknowledged and reset by an operator. This acknowledgment can be configured to happen automatically after a certain amount of time.

Example of a Windows Event Log monitoring:

When a new event is detected, a PATROL event is generated with the exact content of the Windows event so that administrators can easily understand what is wrong and how to solve the problem (the content of the events for Microsoft Lync are notably detailed and provide much information on how to diagnose and troubleshoot the problem). An annotation point is also added to the graph of the MatchingEventCount parameter.

The Windows Event Log sources that are monitored for Microsoft Lync Server are:

• LS A/V Edge Server
• LS ACP MCU
• LS Address Book and Distribution List Expansion Web Service
• LS Address Book Server
• LS AppDomain Host Process
• LS Application Error
• LS Application Server
• LS Applications Module
• LS ApplicationSharing Conferencing Server
• LS Archiving Agent
• LS Archiving Server
• LS Audio-Video Conferencing Server
• LS Audio/Video Authentication Server
• LS Auto Update Server
• LS Bandwidth Policy Service (Authentication)
• LS Bandwidth Policy Service (Core)
• LS Call Detail Recording
• LS Call Park Service
• LS Certificate Manager
• LS Client Version Filter
• LS Common Library
• LS Conferencing Announcement Service
• LS Conferencing Attendant
• LS Configuration Provider
• LS Data MCU
• LS DB Access Layer
• LS Dialin Web Service
• LS Exchange Unified Messaging Routing
• LS File Transfer Agent Service
• LS IM MCU
• LS Inbound Routing
• LS Incoming Federation Service
• LS Intelligent IM Filter
• LS InterCluster Routing
• LS Join Launcher Web Service
• LS LDM
• LS Location Information Service
• LS Lync Web App
• LS Master Replicator Agent Service
• LS MCU Factory
• LS MCU Infrastructure
• LS Mediation Server
• LS Meeting MCU
• LS MGC ADMIN TOOL
• LS MGC CLIENT
• LS MGC COMMON
• LS MGC COMPLIANCE
• LS MGC CONFIG
• LS MGC ENDPOINT
• LS MGC LOADER
• LS MGC LOOKUP
• LS MGC SERVER
• LS MGC SERVICE
• LS MGC TRANSPORT
• LS Outbound Routing
• LS Outgoing Federation Service
• LS Password Expiry Check
• LS Protocol Stack
• LS Provisioning Service
• LS QoE Monitoring Agent
• LS QoE Monitoring Service
• LS Remote PowerShell
• LS Replica Replicator Agent Service
• LS Response Group Service
• LS Routing Data Sync Agent
• LS Script-Only Applications
• LS Server
• LS Software Update Service
• LS Translation Service
• LS User Replicator
• LS User Services
• LS UserPin Service
• LS Web Components Server
• LS Web Conferencing Edge Server

4/Performance Counters

Many performance counters are monitored constantly to report the activity of Microsoft Lync Server and detect potential performance bottlenecks. Each performance counter is represented as a separate instance in the PATROL Console, grouped by the class of the corresponding Windows performance object. Depending on their nature, the performance counter instances may have one or several monitored parameters:

• Value (representing the raw value of the Windows counter)
• Delta (representing the difference of the counter value between two pollings)
• DeltaPerSecond (representing the difference rate of the counter value between two pollings, typically used for bytes/sec or messages/sec counters)

Example of performance monitoring:

The list of monitored performance counters is summarized in the table below:

The following alert thresholds are set by default:

These are the only performance alerts that are generally recommended by Microsoft Lync experts. The other counters are available for further diagnosis when problems occur.

Installation

Prerequisites

In order to setup the monitoring of Microsoft Lync Server 2010, you will need to make sure the following items are available, installed and properly configured:

• A fully functional BMC PATROL environment (optionally part of a larger BPPM environment), with a BMC PATROL Console
• A PATROL Agent on the Lync Server itself
• Monitoring Studio KM for PATROL, installed on the agent on the Lync Server itself, minimum version 8.6.53, with patch P1270
• Monitoring Studio KM is properly loaded on the agent and in the console
• The Lync2012-01-31.cfg installed in a folder on the Lync Server itself

Procedure

1. From the PATROL Console, [right-click] on the main “Monitoring Studio” icon › [KM Commands] › [Configuration] › [Import Configuration…]
   
2. Enter the path of the folder where you have stored the Lync2012-01-31.cfg file (this path is on the agent, on the Lync Server):
   
3. Select the Lync2012-01-31.cfg file in the list:
   
4. Monitoring Studio checks the content of the file. This process can take a few minutes (the configuration file is rather large).
   
5. Monitoring Studio then asks whether the %{PATH} application constant should be cleared. If the default value (“c:\Program Files\Microsoft Lync Server”) happens to match with the installation directory of Microsoft Lync on the server, then you can click on the [Keep values] button. Otherwise, click on the [Clear values].
   
6. Monitoring Studio is ready to import the configuration. Click [Finish] to start the import.
   
7. The import process can take a few minutes:
   
   
   
8. After the import process completes, Monitoring Studio KM starts creating the icons corresponding to the monitored objects in the PATROL Console. If you chose to clear the %{PATH} application constant, all icons are kept OFFLINE and the actual monitoring does not start until you enter a valid value for the application constant:
   
9. To set the %{PATH} application constant value, [right-click] on the “Microsoft Lync Server” icon in the PATROL Console › [KM Commands] › [Modify Application Constants…]
   
10. Next to the %{PATH} constant, enter the path of the folder where Microsoft Lync has been installed on the system and click [OK]. It usually is c:\Program Files\Microsoft Lync Server:
    

Monitoring Studio brings all the monitored objects ONLINE and the monitoring of Microsoft Lync Server starts effectively. The initialization of the monitoring can take a couple minutes to complete.

Alternate Installation Procedure

Alternatively, PATROL administrators can use WPCONFIG.EXE, pconfig or PCM (PATROL Configuration Manager) to deploy the Lync2012-01-31.cfg configuration file. Once this is done, it is recommended to force a full discovery on the PATROL Agent to make sure Monitoring Studio KM takes into account the new configuration immediately (without waiting for the next discovery cycle, which occurs by default every hour).

Editing the %{PATH} application constant as described above is still required to make sure the monitoring is adapted to the environment. This can also be done by editing the /MASAI/SENTRY8/Lync/constant1Value configuration variable before applying the configuration to the selected Lync servers.

Post-installation tasks

As explained earlier, depending on the role of a Microsoft Lync Server, different components of Lync Server have been installed and configured. The monitoring configured in Monitoring Studio covers all components of Lync and you may need to disable or completely remove certain groups of monitored objects from the monitoring.

The different aspects of Microsoft Lync are grouped in containers/folders in Monitoring Studio. In order to identify the components that would need to be removed from the monitoring, you simply have to first “browse” the tree view to find “Windows Service” objects for which the Status parameter cannot be collected (it stays OFFLINE, while the rest of the monitoring is properly collected). Also, an error message is displayed in the System Output Window for each service that is not installed and thus whose monitoring cannot be performed:

For each “container” with a Windows service that is not installed and cannot be collected, you can remove it from the monitoring. [Right-click] on the container icon › [KM Commands] › [Delete]:

Once this is done, you should no longer see any error message in the System Output Window and the monitoring is adapted to your environment.

Related Topics

• Monitoring Microsoft Lync Server 2013/Skype for Business with Monitoring Studio 8 or 9
• Monitoring Proxmox Systems using Monitoring Studio KM
• Monitoring Squid Cache Proxy using Monitoring Studio KM
• Monitoring Dell EMC Avamar Appliances using Monitoring Studio KM
• Monitoring NetApp E Series Storage Devices using Monitoring Studio KM