Answer by Andreas G. ·
The number of deployed sensors should not impact whehter your JVM crashes or not. If you really experience crashes I suggest you open a support ticket. It "should not" have to do with the number of deployed sensors
Answer by Andreas G. ·
Hi Andrew
They key metric is not necessarily the number of deployed sensors but the size of your PurePaths. If you look at the Deployment Guide you will see that one of the key metrics is "Nodes per PurePath". In Step 1 - Collect Sizing Data it says that an average PurePath in production environment should have between 10 and 50 nodes. In a testing environment it can be higher. Of course - all of this depends on the number of transactions you process in a given time interval.
The general Best Practice for custom sensors is to only instrument those methods that give you business or root cause context, e.g: methods where you need to capture arguments or return values or methods of critical APIs or Interfaces that you need to monitor for SLA reasons
Andi
I ask because one of our tier's jvms crash when we add dynatrace agents. To date we have not been able to figure out the conflict. I noticed that most of our agents have ~5k sensors deployed but the problematic agents have ~40k. Seems like a significant difference.
None of the purepaths have node sizes greater than 300. And 1% of the sample I took are over 50 nodes (per the size column in the purepath dashboard). So those numbers seem close to guidance. The documentation mentions an average node size. Is that a metric dynatrace calculates for me that I view?
JANUARY 15, 3:00 PM GMT / 10:00 AM ET