Key Monitoring Scenarios

更新日期: 2010年6月

This topic describes the key monitoring scenarios for the Windows Server AppFabric Management Pack.

AppFabric Databases

AppFabric uses two types of databases to provide application monitoring and workflow persistence capabilities. The persistence database is critical to the operation of the workflows that are running in the AppFabric/IIS server computer. The monitoring database is not as critical, because the application logic does not depend on its data. However, if the monitoring database becomes unavailable, users will lose visibility into what the applications are doing.

This management pack monitors these runtime databases from the perspective of the AppFabric server computer. That means that the monitors actually run on the computers that have AppFabric installed and not on the computer where the database is installed. For that sort of database monitoring, refer to the SQL Server Management Pack. The database monitors use the following two techniques to determine the health of an AppFabric database:

  1. The management pack checks the local AppFabric event log to see if the AppFabric runtime is reporting errors that indicate database connectivity problems.

  2. The management pack actually queries the database to check for typical problems.

AppFabric Runtime Infrastructure

AppFabric Hosting Services provides runtime capabilities that help customers manage and monitor Windows Communication Foundation (WCF) and Windows Workflow Foundation (WF) services. There are three runtime services that provide these capabilities:

  • Workflow Management Service

  • Event Collection Service

  • Cache Host Service

The management pack provides monitors that determine the health of these three services. The event log is the primary source of health indicators.

AppFabric Managed Applications

AppFabric provides monitoring features for WCF and workflow services. This management pack acts as a proxy that forwards a subset of that monitoring information from the AppFabric system into Operations Manager. This allows you to get basic insight into the performance of your WCF services without having to use the AppFabric user interface. The management pack exposes the following WCF metrics and allows alerting thresholds to be configured on them:

  1. WCF Operation Calls Per Second

  2. WCF Average Operation Call Duration (milliseconds)

  3. WCF Operation Calls Failed Per Hour

Monitoring Time Intervals

The following groups of monitors should always be configured to use the same interval and timeout values in order to work properly. This will never be an issue if you are using the sealed MP. However, if you unseal the MP and modify it, you will need to keep this in mind.

AppFabric SQL Monitoring Database Monitors

  • Operation History Failure Count

  • Failed Staging Table Row Count

  • SQL Instance Store Co-Location

  • Staging Table Row Count

  • Staging Table Transfer Events Job

  • Staging Table Wcf Events Job

  • Staging Table WF Events job

AppFabric IIS Web Service Monitors

  • Average Call Duration

  • Operation Call Rate

  • Operation Failure Rate

  • Rule - Web Service Operation Failure Rate Collection

AppFabric Cache Monitors

  • HA Enabled Secondary Node Down

  • Named Cache Not Accessible

AppFabric Cache Management Installation Monitors

  • Cluster Config Store State

  • Cluster Lead Host State

  • Cluster State

Monitors That Have Manual Reset States

The following monitors cannot be reset using a timer or any other automated mechanism. When one of these monitors is triggered, and you resolve the issue, you must then manually reset the health of the target object.

 

Target Monitor Description

AppFabric Event Collection Service

Register Trace Provider State

This event indicates that the Event Collector failed to add a particular event source (provider) to the current session. Monitoring will not work for an application that is a part of this source.

AppFabric Event Collection Service

Provider in Too Many Sessions State

This event indicates that the Event Collector failed to add a particular event source (provider) to the current session. Monitoring will not work for an application that is a part of this source.

AppFabric Event Collection Service

Read application config - General failure

This event indicates that the Event Collection Service was unable to read a particular web.config file.

AppFabric Event Collection Service

Unregister Trace Provider State

This event indicates that the Event Collection Service could not unregister an event source (provider) from an ETW session.

AppFabric Event Collection Service

App Connection String Reference State

This event indicates that a particular IIS web.config file has enabled monitoring, and specifies a connection string reference. However, the referenced connection string could not be found.

AppFabric Event Collection Service

App Unique Provider Id State

This event indicates that multiple IIS Web applications have configured monitoring at different levels, but use the same provider Id.

AppFabric Event Collection Service

App Config Readable State

This event indicates that the Event Collection Service encountered an unhandled exception that was unexpected.

AppFabric Workflow Management Service

net.pipe protocol is not activated

This event indicates that a particular web site or application does not have the net.pipe protocol enabled. This protocol is required for the Workflow Management Service to communicate with service hosts.

AppFabric Workflow Management Service

Unknown Argument Supplied

This event indicates that an unknown startup argument has been passed to the Workflow Management Service.

社群新增項目

顯示: