Datastore Capacity Exhausted Due to Virtual Machine Swap File Location Policy Override
search cancel

Datastore Capacity Exhausted Due to Virtual Machine Swap File Location Policy Override

book

Article ID: 435416

calendar_today

Updated On:

Products

VMware vCenter Server VMware vSphere ESXi

Issue/Introduction

VMware vCenter Server and VMware ESXi. Datastore capacity exhaustion occurs when virtual machine swap files (.vswp) are created in the virtual machine home directory instead of the designated host-configured swap datastore.

The hostd log indicates the local policy is disabled:

2026-01-21T22:31:17.961Z In(166) Hostd[<REDACTED_ID>]: [Originator@6876 sub=Solo.ComputeResource opID=HB-SpecSync-host-<REDACTED_ID>@<REDACTED_ID> sid=<REDACTED_ID> user=vpxuser:<no user>] **Updated local swap datastore policy: false**

The vpxd log shows a cluster reconfiguration task prior to the host update:

2026-01-21T22:31:12.502Z info vpxd[<REDACTED_ID>] [Originator@6876 sub=vpxLro opID=<REDACTED_ID>] [VpxLRO] -- BEGIN task-<REDACTED_ID> -- domain-c<REDACTED_ID> -- **vim.ComputeResource.reconfigureEx**

Environment

VMware vCenter Server 8.x
VMware vCenter Server 9.x
VMware ESXi 8.x
VMware ESXi 9.x

Cause

The primary cause is a cluster-level reconfiguration that changed the "Swap File Location" setting from "Host configured location" to "Virtual Machine Directory". This override forces all hosts within the cluster to store swap files on the same datastore as the virtual machine configuration files, overriding local configurations and leading to capacity exhaustion on the affected datastores.

Resolution

  1. Navigate to the vSphere Cluster settings in the vSphere Client.

  2. Go to Cluster > Configure > General > Swap File Location.

  3. Ensure Host configured location is selected.

  4. Verify each ESXi host has the correct local swap datastore specified under Host > Configure > Virtual Machines > Swap File Location.

  5. Power cycle (power off and power on) the virtual machines for the new swap file location to take effect. A guest OS reboot is not sufficient.

This logging issue regarding the visibility of the exact reconfigureEx attribute modification is under review with Broadcom Engineering. A change has been made in code to make the logging more verbose in vpxd.log by default.