ReSearch This KB – APM Data Transfer Health

Alert: APM Data Transfer Health

Management Pack Name: Operations Manager APM Infrastructure

Management Pack Version: 7.1.10226.0 (or probably any version)

Rule or Monitor? Monitor

Rule or Monitor Name: APM Data Transfer Health

Rule or Monitor Notes

The instance state becomes unhealthy when an error is detected on the Management Server when it is performing Data Transfer, Aggregation, and Grooming activities related to APM Events on the Operations and/or Data Warehouse databases.

Typically, this depends on availability of the database, its load, and general load on the system. It could be a transient condition in which the monitor will reset to healthy state after one hour.

Verify that the Operations and Data Warehouse Databases are up and running. If this happens frequently, consider reducing the amount of data collected by APM by tuning or adjusting thresholds.

Issue: You are seeing event 34505 with source .NET app monitoring in Operations Manager log on one of the Management Servers, repeated every five minutes.

An old date is returned when running this query against the Data Warehouse Database:

use [OperationsManagerDW]

SELECT TOP (1) startDate FROM apm.tasks (NOLOCK)

Causes: The APM workflow Microsoft.SystemCenter.Apm.DataTransferRule will run the above query and check the result against a threshold value of three hours (DatabaseIsNotSyncMoreHours). It this threshold is breached the monitor will change to a Warning state.

Resolution: Old startDate could indicate that the synchronization do not run properly. It can also indicate that you have old tasks that belong to a Management Group no longer in use. To see all APM tasks, run this query:

select * from apm.tasks (nolock)

If you see SEVIEWERDBID with value 1 and old dates, and SEVIEWERDBID with value 2 and recent dates, you probably have APM tasks from an old Management Group.

Manually updating Operations Manager databases is not supported by Microsoft. You should contact Microsoft Support for assistance if this happens in a production environment.

The solution will probably be to run this delete query against the Data Warehouse database (remember, this is totally unsupported if you do it on your own initiative):

USE [OperationsManagerDW]

DELETE FROM apm.tasks

 

This article was written by Reidar Johansen

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.