Hi,
in the course of Januari we extended the instrumentation of our tomcat production servers with AD agents.
A subset of the servers was already monitored for almost a year. The extension made the coverage complete over all servers.
The full set has been active in production for a couple of weeks already.
This morning a few servers (two types) displayed high cpu and became unresponsive.
Thread dumps taken (attached) showed a simlar class, unknown to our code, to be the hot spot.
Similar reports linking these classes to AppDynamics
(e.g.
http://stackoverflow.com/questions/23276678/java-memory-leak-with-concurrentlinkedqueuenode ) made our support turn off the agents, which effectively deblociked the situation.
The two types of servers ( two independent production parts) do both interact with an ActiveMQ server ( also each with a independent, separate ActiveMQ server).
Are similar situations known where the agent negatively influences the monitored server and what is the actual mechanism behind ?
Is the indicated class interacting with the AppDynamics agent ?
Regards,
Hi Ralph,
Can you check now, the links:
http://help.appdyna
Or http://help.appdyna