Disabling "Automatic Transaction Tracing" to avoid high overhead.

Products

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

Issue/Introduction

  In a customer's production environment, they are seeing excessive traces being generated. They would like to turn off the "Automatic Transaction Tracing" to avoid high overhead. 

They are observing the following errors from agent logs every 15 seconds. 

"The Transaction Tracer reporting limit of 200 traces...

 

 Do they have to set the following flags to false to disable header injection?

#Headers 
com.wily.introscope.agent.soapheaderinsertion.enabled=false 
com.wily.introscope.agent.soapheaderread.enabled=false 

com.wily.introscope.agent.httpheaderinsertion.enabled=false 
com.wily.introscope.agent.httpheaderread.enabled=false 

#disable MQ correlation 
#com.wily.introscope.agent.mqheaderread.enabled=false 
#com.wily.introscope.agent.mqheaderinsertion.enabled=false 

 

Environment

In any 10.x release. This was asked for a 10.3 environment.

Resolution

The six properties are not related to the issue and will not disable the message. They are used to correlate traces.  In any case, traces were generated and there is no point knowing whether they were correlated or not.

 

Determine 

- What kind of traces are flooding? SAMPLED, NORMAL, AUTO TRACES. 

- Were they correlated. Are all the agents flooding traces or a few? 

- Are there Any .NET agents?

 

Sampled traces in upstream JVMs can trigger multiple traces in downstream JVMs.

So setting the property introscope.agent.transactiontracer.tailfilterPropagate.enable to false and disabling sampling in all the upstream JVMs can prevent multiple traces in downstream JVMs.  This is a hot property.  In case if it does not work, you can restart the app.

AutoTrace is also a NORMAL trace but AutoTrace will have AutoTrace trigger criteria.

 

If you are seeing multiple NORMAL traces,  then check whether they are due to AutoTrace feature. The trace contains AutoTrace Trigger criteria if it is an AutoTrace.  In 10.5, AutoTraces can also be propagated to downstream JVMs'.

 

If traces are due to the AutoTrace feature, then you can turn off autotracing at the EM. But the thresholds that were set already at agent will not be cleared until agent is also restarted.

 

Note that all of the properties discussed above are applicable only to agent.

Additional Information

TEC1895963 -- Disabling SOAP/HTTP header injection for Java and .NET Agents