Linux Kernel CPU Profiling

the linux kernel cpu profiler can be enabled in observe agent to collect machine level cpu statistics our profiler internally leverages ebpf to collect stats so enabling this adds minimal amount of latency on your overall system since the profiler internally uses ebpf it needs your cluster to be running with a linux kernel version >= 4 9 getting started ensure that observenow profiling is enabled for your stack this sets up the profiling server along with a username and password to login login to the profiling server with your credentials, and then navigate to settings > api keys > add api key and create a new api key with following config name observeagentkey role agent valid for 0 # no expiry make sure you copy the api key created and add the following yaml snippet to the agent's values yaml and run the agent based on these instructions https //github com/devopsnowinc/docs/blob/opsverse/docs/observenow/install opsverse agent/#kubernetes pyroscope ebpf enabled true args \ "ebpf" \ " application name" \ "agent ebpf" \ " server address" \ "\<add server endpoint here>" \ " auth token" \ "\<add apikey here>" you should see kernel level cpu statistics start getting pushed to your profiling server within a few minutes for more pyroscope options, visit this repo https //github com/pyroscope io/pyroscope