We are using Dynatrace to monitor an application on the AWS environment. A few EC2 nodes being monitored are standalone JVMs and do not have a web appserver associated with them. We have manually configured sensors for them so that they show on the transaction flow and we are able to access the related PurePaths.
For one of the scenarios, per the autoscaling policy of these nodes, 2 new nodes get added mid execution. The new nodes are recognised by Dynatrace and show up in the transaction flow. But when we drill down to view the PurePaths for these autoscaled nodes, there are quite a few missing PurePaths, (as compared to the original node on which the execution had started) especially the PurePaths having methods interacting with other nodes. All the PurePaths that show up for the autoscaled nodes are related to code internal to the node, none of them showing methods that talk to the other nodes in the environment.
On seeing the 'Deployed Sensors' for the agent of the autoscaled node, our manually created sensors can be viewed. So the sensors are being picked up correctly but due to some reason, a few PurePaths are missing.
Any special considerations that we need to take into account for monitoring nodes that have auto-scaling possibility?
Appreciate the help.
Your Auto-Scaling nodes should not be any different than your regular nodes. I assume you have configured the Dynatrace Agent the same way as for your regular nodes - meaning - using the same Agent Name which means they map into the same agent group.
Have you tried to look into the Agent Log files for these auto-scaling nodes? would be interesting if soemthing happens during startup where not all code gets instrumented.