Aria Operations Telegraf metric collection interval increases from 5 minutes to 10 minutes

Products

• VCF Operations

Issue/Introduction

Symptoms:

• Metric collection for Telegraf agents in the VMware Aria Operations UI shows a delay of 10 minutes (or longer) despite being configured for a 5-minute interval.
  
  
• Data points in dashboards and reports appear at every second collection cycle.
  
  
• The environment has recently seen an increase in the number of monitored objects or the addition of new Telegraf plugins/scripts.
  
  
• The environment utilizes a single Cloud Proxy (CP) for a large number of Telegraf agents (e.g., 1,500+).

Environment

• VMware Aria Operations 8.18.x
  
  
• Telegraf Open Source Agents

Cause

The data collection cycle is exceeding the default 5-minute (300 seconds) window. This typically occurs when the collection layer is undersized for the number of monitored objects. When the Cloud Proxy cannot complete the collection for all agents within the 5-minute window, data points are calculated and reported only every second cycle, resulting in a 10-minute visible delay in the UI.

In observed cases, doubling the monitored objects (2X increase) without adding additional Cloud Proxies has led to this architectural bottleneck. 

Resolution

To resolve the metric delay, you must offload the collection workload to ensure cycles complete within 300 seconds.

1. Upgrade Aria Operations

• Upgrade to Aria Operations 8.18.6 or later. This version includes optimizations for faster metric processing.
• Alternatively, upgrade to VMware Cloud Foundation (VCF) 9.1, which introduces Load Balanced Processing of Telegraf metrics across the cluster.

2. Deploy Additional Cloud Proxies

• Deploy one or more additional Cloud Proxies to distribute the Telegraf agent load.
• Ensure Cloud Proxies use the Standard sizing (4 vCPU and 32GB RAM).
• Reconfigure a portion of the Telegraf agents to report to the newly deployed Cloud Proxy node.

3. Verify Sizing and Health

• Navigate to Administration > Cluster Management and check the Nodes Object Load (%).
• The object load should ideally remain below 80% for stable collection.
• If the load remains high after adding a CP, consider deploying a Collector Group to leverage high availability and further load distribution.

If the above steps fails to resolve the issue then apply proper HA proxy configuration to balance the load.

Additional Information

Aria Operations Cloud Proxy becomes unresponsive

VMware Aria Operations 8.14 Sizing Guidelines