Cloud Proxy Memory Exhaustion and Collection Failures in Aria Operations 8.18.x Due to Unfiltered NSX BGP Metrics

Products

• VMware Aria Operations (formerly vRealize Operations) 8.x

Issue/Introduction

In VMware Aria Operations 8.18.x, after starting the NSX Management Pack adapter instance, the Cloud Proxy experiences memory crashes, collection failures, or silent adapter restarts. While the first configuration collection cycle succeeds, subsequent performance collection cycles are silently interrupted before completing or writing standard exception logs.

Environment

VMware Aria Operations 8.18.x

Cause

A code defect in the NSX Adapter 8.18.x branch causes BGP-related metrics (BGPNeighborInstancedMetric) to bypass the AdapterBase#shouldCollect filtering logic during ingestion. Even if BGP metrics are disabled in the monitoring policy, the adapter fetches all raw BGP route and neighbor metrics into memory before evaluating the policy rules. In environments with a large volume of BGP routes, this behavior leads to extreme heap memory bloat ($90\%+$ of total heap consumed) and triggers Out-Of-Memory (OOM) crashes on the Cloud Proxy.

Resolution

Applies metric filtering at the API ingestion level before data is loaded into memory.

This issue will be resolved in an upcoming version.

Additional Information

After applying the patch, if users experience difficulties locating specific NSX alarms within vROps, verify object mappings in the UI. In vROps, NSX alarms are distributed across their corresponding parent objects (e.g., NSX Adapter Instance, Transport Node, Router Service) rather than consolidated into a single alert view.