Hardware Health showing “System Memory Non-Recoverable” status for ESXi hosts in NPM

When monitoring VMware ESXi hosts in NPM, the Hardware Health resource on an ESXi node can show the System Memory sensor in a Non Recoverable or critical state, even though the host appears healthy in tools such as iLO or the vCenter UI and the host is still being successfully polled by NPM.

NPM does not independently test ESXi hardware; instead, it displays the hardware status that VMware exposes via its hardware-health interfaces (CIM / VMware API). For ESXi hosts polled through vCenter or directly via VMware polling, the System Memory sensor status in NPM is taken from the ESXi host’s hardware health data (for example, hardwareStatusInfo > memoryStatusInfo in the VMware Managed Object Browser, or equivalent CIM data).

If VMware reports the memory sensor in a red / Non Recoverable state, NPM marks the System Memory sensor as critical, even if other vendor tools (such as iLO) appear healthy. In some cases, the VMware-side sensor can become stale or inconsistent and needs to be validated and refreshed.

SolarWinds is most likely displaying the memory status returned by the ESXi/VMware hardware-health API, even if iLO or the vCenter UI looks healthy.

Verify whether the host is polled through vCenter or directly through ESXi, validate the same memory sensor via the ESXi API/CIM, and then disable/re-enable the Hardware Health resource from List Resources on the affected node. 

Disable and re-enable hardware health sensors from the list resource.

To do this, go to:

1. ESXi2 host page > click on list resources > and disable the Hardware Health Sensors, then select Submit: 
 
2. After disabling, click on list resources again and re-enable, then select Submit.