False alerts from agents swapping collectors

book

Article ID: 209755

calendar_today

Updated On:

Products

CA Application Performance Management (APM / Wily / Introscope)

Issue/Introduction

 
 

There are false alerts in APM 10.7 workstation due to the agents are swapping from one collector to another one.

There is a moment when that agent is not aware of the new collector and it triggers an alert causing false alerts and creating confusion in the production environment.

There is a desperate need to know how to fix this as we are using Introscope to monitor when a server is offline.

Cause

 
 

The default Connection Status metrics are not cluster-aware. So if an agent disconnects from one collector and immediately moves to another collector, it will still show disconnected (value=3) in relation to the first collector but connected (value=1) to the second collector. The alert is thrown because the threshold is set to fire above a value of 2 (represents an agent reporting slowly)

Environment

Release : 10.7.0

Component : APM Agents

Resolution

There is a Javascript calculator which provides a cluster-aware version of the connection status metric. This can be copied into the scripts folder of the MOM

It will provide a metric in this location:

*SuperDomain*|Custom Metric Host (Virtual)|Custom Metric Process (Virtual)|Custom Metric Agent (Virtual)|Agents|<agent name>:Clusterwide Agent Connection Status

Additional Information

 

Taken from APM community where there is more background discussion of the script:

https://community.broadcom.com/enterprisesoftware/communities/community-home/librarydocuments/viewdocument?DocumentKey=c57e783b-1e32-4e02-944c-1e2cf17d3d70&CommunityKey=be08e336-5d32-4176-96fe-a778ffe72115&tab=librarydocuments

Attachments

231149455_1614872119949.zip get_app