High CPU usage at regular intervals on an Aria Operations node

Products

• VMware Aria Operations 8.x

Issue/Introduction

The Aria Operations node started to have CPU spikes at regular intervals, which could happen after restarting a node or after an upgrade. The collection process might still be functioning correctly.

Cause

This can occur because the collection cycles for most adapter instances are aligned, leading to resource strain (e.g. Kubernetes Adapter Instances can require resources), particularly when data collection is initiated simultaneously. After a node restart or an upgrade process, the adapter instances might start at the same time.
This situation is rare.

Resolution

Stop some Adapter Instances and restart them after a delay of 1 or 2 minutes. This action will disrupt the aligned collection cycles.

Additional Information

To check the collection cycle alignment, inspect the “Instance Generated|Data Collection Duration (ms)” metric of all the Adapter Instances of the problematic node. Hover over the graph to get the timestamp and collection duration. If they are similar between the Instances, it will confirm the alignment. (See example in screenshot below)