"Baseline prediction not available for" messages filling up MOM log quickly.
search cancel

"Baseline prediction not available for" messages filling up MOM log quickly.

book

Article ID: 36995

calendar_today

Updated On:

Products

CA Application Performance Management Agent (APM / Wily / Introscope) INTROSCOPE

Issue/Introduction

  Enterprise Manager(Collector and MOM) outputs the below warning message every 15 seconds.

  [WARN] [pool-<n>-thread-<m>] [Manager] ATDS: baseline prediction not available for: SuperDomain|<Host_Name>|<Process_Name>|<Agent_Name>|Frontends|Apps|<App_Name>|URLs|Default:Average Response Time (ms)
 
  e.g.:
  [WARN] [pool-10-thread-18] [Manager] ATDS: baseline prediction not available for: SuperDomain/APS EXT|sbr00ap03|WebSphere|APPECell01/APPECluster02.Srv04|Frontends|Apps|BR0573C08_ConfirmacaoPresenca|URLs|Default:Average Response Time (ms)

Environment

CA Application Performance Management 10.7 release.

Cause

  This is due to a new feature added to APM 10.x - AutoTrace with Differential Analysis. 

  Configuring alerts and thresholds are too time consuming, even with “automated” baselines which are too noisy or require a lot of configuration.
  By using simple statistical techniques, CA APM Performance Spike Detection discovers spikes in application performance, proactively triggers evidence (trace) collection, and displays a rich performance story across applications, all with simple configuration that works out-of-the-box.

Resolution

  Option 1:

     The warning message will stop after the Agent sends the metric which be output in the warning message.
 Set the following property in IntroscopeEnterpriseManager.properties
(Collector and MOM) if you do not need Differential Analysis data. This will then stop the message.

 introscope.enterprisemanager.baseline.max.metrics.baselined=0

  Option 2:
 
    You could set the logging level to "errors." Then you won't get any warning messages. Make this change under IntroscopeEnterpriseManager.properties (Collector and MOM) as below:
 
    log4j.logger.Manager=ERROR, logfile
 
  Option 3:
 
    If you think you are getting too many notifications, then you could try to reduce notifications for a front end.