Why is hypervisor disk_available_least greater than free_disk_gb?

From the documentation I've found, disk_available_least should always be less than free_disk_gb.

Example:

5000G host filesystem
400G actually used by instances
4000G virtual size of instances

free_gb = 5000 - 400 = 4600
overcommited_total_gb = 4000 - 400 = 3600
disk_available_least = free_gb - overcommitted_total_gb = 4600 - 3600 = 1000

---
However, I have a hypervisor in the following state:
disk_allocation_ratio = 1.5

> openstack hypervisor show hostname-stripped
+----------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Field | Value |
+----------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| aggregates | [] |
| cpu_info | {"vendor": "Intel", "model": "SandyBridge", "arch": "x86_64", "features": ["pge", "avx", "vmx", "clflush", "sep", "syscall", "vme", "dtes64", "tsc", "xsave", "xsaveopt", "xtpr", |
| | "cmov", "ssse3", "est", "pat", "monitor", "smx", "pcid", "lm", "msr", "nx", "fxsr", "tm", "sse4.1", "pae", "sse4.2", "pclmuldq", "acpi", "tsc-deadline", "mmx", "osxsave", "cx8", "mce", |
| | "de", "tm2", "ht", "dca", "pni", "pdcm", "mca", "pdpe1gb", "apic", "sse", "pse", "ds", "invtsc", "lahf_lm", "aes", "sse2", "ss", "ds_cpl", "arat", "pbe", "fpu", "cx16", "pse36", |
| | "mtrr", "popcnt", "rdtscp", "x2apic"], "topology": {"cores": 8, "cells": 2, "threads": 2, "sockets": 1}} |
| current_workload | 0 |
| disk_available_least | 89 |
| free_disk_gb | 31 |
| free_ram_mb | 5586 |
| host_ip | 172.17.37.7 |
| host_time | 16:22:24 |
| hypervisor_hostname | hostname-stripped |
| hypervisor_type | QEMU |
| hypervisor_version | 2006000 |
| id | 7 |
| load_average | 7.85, 8.02, 7.68 |
| local_gb | 771 |
| local_gb_used | 222 |
| memory_mb | 131026 |
| memory_mb_used | 66270 |
| running_vms | 13 |
| service_host | hostname-stripped |
| service_id | 16 |
| state | up |
| status | enabled |
| uptime | 44 days, 37 min |
| users | 0 |
| vcpus | 32 |
| vcpus_used | 44 |
+----------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+

To simplify the problem, I have only included one hypervisor. However, all 6 of my hypervisors are currently in this state. I am using kolla, so perhaps a mass-misconfigured something.

Why would a hypervisor be in this state?