Nutanix: Fix incorrect memory usage metrics for VMs

6 votes

Information from Nutanix support:

> I’ve been running several tests using the v1 API call on a test VM across different clusters, and I’ve found the following:
> + The API being used returns different representations of VM memory, including guest.memory_usage_ppm, guest.memory_usage_bytes, and memory_usage_ppm.
> + The value shown in the Prism UI is derived from memory_usage_ppm, which reflects memory usage from the hypervisor’s QEMU perspective. This is considered more accurate.
> + On the other hand, guest.memory_usage_bytes or _ppm represent memory usage from the VM’s internal perspective, which is less accurate as it doesn’t account for memory used by the OS for filesystem caches and buffers.

Please use the more accurate request: memory_usage_ppm.

Under consideration Checks&Agents Suggested by: defdefred (13 Nov, '25) • Upvoted: 31 Jan • Comments: 0