Dell PowerStore Storage Exhaustion with vSphere Replication

Products

• VMware vCenter Server

Issue/Introduction

Storage arrays, specifically Dell PowerStore, report critical physical capacity utilization (e.g., >98% usage) even when vCenter shows datastores are only partially consumed. Common symptoms include:

• Dell PowerStore Event Code 0x00200e07 (Physical capacity utilization above 98 percent).
• System impact warnings: Cluster performance impacted, or upgrades not allowed until resolved.
• Discrepancy between "Used Space" in vCenter and "Physical Used Space" on the storage array.

Environment

• vSphere 8.x
• vCenter Server 8.x
• vSphere Replication

Cause

This discrepancy can occur in environments utilizing vSphere Replication with Point-in-Time (PIT) replicas, combined with thin-provisioned storage.

• vSphere Replication Overhead: Replication data (initial full sync + PIT delta redo logs) consumes significant physical space that is not always registered as part of a VM's used space in the vSphere GUI.
• Overprovisioning: Datastores presented to ESXi as thin-provisioned volumes may not be fully backed by physical storage on the array.
• Unused Block Accumulation: Deleted data within the Guest OS or VMFS may not be automatically reclaimed on the physical array without specific UNMAP configurations.

Resolution

1. Manual Space Reclamation (UNMAP)

Trigger manual reclamation of deleted VMFS blocks to the storage array:

• • Run the manual unmap command: esxcli storage vmfs unmap -l <Datastore_Label>.
    • See TechDocs: Space Reclamation on vSphere VMFS Datastores
  • Verify that VAAI (specifically the "unmap" function) is enabled on the ESXi hosts.
  • Monitor the host to ensure unmap commands are processed without failure.
    
    • See KB: Monitor automatic unmap I/O issued by ESXi

 

2. Guest-Level Space Reclamation

Ensure the Guest Operating System is configured to reclaim space from within the virtual disk:

• • Verify that TRIM/UNMAP is enabled inside the guest OS to allow the filesystem to notify the hypervisor of deleted blocks.
    
    • See related KB: Procedure to Enable TRIM/UNMAP to reclaim space

 

3. Optimize vSphere Replication

If vSphere Replication is the primary consumer of space:

• • Reduce PIT Instances: Lower the number of retained Point-in-Time (MPIT) instances in the replication configuration
    
    • Change the Point in Time Settings of a Replication
      
      
      
  • Cleanup Stale Data: Remove snapshots and any old virtual machines that are no longer required.

 

4. Physical Storage Expansion

If reclamation and optimization do not provide sufficient relief, physical capacity is exhausted due to overprovisioning.

• Add additional physical disks to the storage pool or move storage resources within the cluster to another appliance as recommended by the array vendor.

Additional Information