DX SaaS APM : UMA Teams Consuming High CPU and Memory

Products

• DX APM SaaS

Issue/Introduction

UMA Teams Consuming High CPU and Memory

Environment

DX APM SaaS

System Edge 

UMA Agent with host monitor Extension

Open Shift  

Cause

Most likely it's the SNMP Collector thread pool got overloaded, as we poll each node every 15 seconds

Resolution

Workaround: the result of polling interval config change last time is probably the best what we can get at the moment

 to reduce the overall CPU usage is to increased the polling interval (introscope.agent.hostmonitor.profiles.localhost.interval)

Please try add this env. variable to increase the polling interval to 5 minute:

       - name: apmenv_introscope_agent_hostmonitor_profiles_localhost_interval
         value: "300"
 

Another thing to try is to reduce the SystemEdge metrics by editing this env. variable to remove groups not needed:

        - name: apmenv_introscope_agent_hostmonitor_profiles_localhost_linux_groups
         value: memory,cpu,network,diskio,hostresources,application,protocol,systemmanagement,performance,process,ipc,distributedsystem,topprocess,operatingsystem,storage,systemload,port