| Previous | Next |
| ERROR_NDIS_DOT11_AP_BAND_NOT_ALLOWED | WHV_E_UNKNOWN_CAPABILITY |
ERROR_VID_REMOTE_NODE_PARENT_GPA_PAGES_USED
ERROR_VID_REMOTE_NODE_PARENT_GPA_PAGES_USED is HRESULT 0x80370001 in the NUMA placement advisory area of the Windows virtualization stack. It applies to host-side placement of a VM's guest-physical backing pages across more than one physical NUMA node.
What the code establishes
This is a placement warning, not evidence that guest memory is corrupt. It records that at least part of the VM working set is remote from the processor locality that owns or frequently accesses it.
Neighboring result: Contrast it with invalid NUMA configuration: this status can accompany a running VM and matters primarily when latency, bandwidth, or cross-node traffic is measurably poor.
Evidence that should survive remediation
| Evidence | Why it changes the diagnosis |
|---|---|
| VM and host NUMA topology | Record vCPU count, virtual NUMA nodes, assigned memory, host node sizes, and whether NUMA spanning is enabled. |
| Placement and pressure | Capture host free memory by node, Dynamic Memory state, startup RAM, and whether another VM forced remote allocation. |
| Workload symptom | Keep CPU-ready time, memory latency indicators, and the exact interval in which throughput or tail latency changed. |
| Mobility history | Note live migration, restore, host reboot, or configuration changes that preceded the advisory. |
Controlled diagnostic sequence
- Compare performance with the VM placed on a host that can satisfy its working set within one NUMA node.
- Keep guest workload, vCPU count, and memory size fixed while changing only NUMA placement or spanning policy.
- Check whether the warning remains when competing VMs are stopped or moved.
- Do not delete saved state or virtual disks merely because this informational result appears.
Evaluate the condition with the same workload and VM size while varying only documented NUMA configuration or placement. Record topology on every intended host rather than inferring it from socket labels.
Read the comparison tests
| Control | Interpretation | Hold constant |
|---|---|---|
| Automatic versus explicit virtual NUMA | If automatic topology removes it, compare each overridden node count, processor limit, and memory limit with host topology. | Keep VM vCPU count, memory, Dynamic Memory mode, and workload fixed. |
| Every intended destination host | A result that follows one node indicates a host-local topology or placement constraint. | Use the same VM configuration and record each host NUMA layout in the test. |
| Local versus cross-node placement | Measure workload latency and throughput under both placements. This distinguishes a correctness failure from a performance advisory. | Hold guest workload and processor affinity stable while comparing it. |
Avoid the wrong recovery
A restart can change placement and make it disappear once without proving the topology is valid or performance is stable. Close the investigation only after repeatable placement and workload comparisons.
Acceptance test
The condition is understood when either performance remains acceptable despite remote pages, or a controlled locality change removes both the advisory and the measured regression.
Technical references
- Microsoft Open Specifications: HRESULT values — used to interpret this result.
- Microsoft: NUMA and Hyper-V virtual machines — used to interpret this result.
- Microsoft: configure virtual NUMA — used to interpret this result.
- Microsoft: troubleshoot Hyper-V VM performance — used to interpret this result.
Looking for a different code? Search another status or error code.