VM Restore and vMotion Failures due to APD Events.
search cancel

VM Restore and vMotion Failures due to APD Events.

book

Article ID: 434678

calendar_today

Updated On:

Products

VMware vCenter Server

Issue/Introduction

Following a VM restore from a third-party backup solution to vCenter, the migration (vMotion) from the backup storage to production storage fails. This often leads to All Paths Down (APD) conditions or transient storage connectivity issues.

 

  • The vMotion task fails or hangs indefinitely.
  • The VM becomes "Inaccessible" or shows as "Orphaned" in the vSphere Client.
  • Network adapters for the affected VMs appear disconnected after the restore.
  • Storage latency or APD events recorded in the ESXi vmkernel.log.

Environment

VMware vSphere / ESXi

Cause

The APD events and subsequent vMotion failures are caused by an MTU mismatch between the ESXi hosts and the third-party backup appliance.

While the ESXi VMkernel interfaces are configured for Jumbo Frames (9000), the backup appliance is set to standard frames (1500). When the ESXi host attempts to transfer large data blocks during the vMotion or restore process, the packets are dropped or fragmented. This causes the ESXi host to lose connectivity to the NFS datastore, resulting in an APD condition.

Resolution

To resolve these APD events, you must synchronize the MTU settings across the entire storage network path.

  1. Check ESXi MTU:

    • In the vSphere Client, select the ESXi host.

    • Navigate to Configure > Networking > VMkernel adapters.

    • Ensure the MTU for the storage/NFS adapter is set correctly (e.g., 9000).

  2. Verify Network Path:

    • Confirm that all physical switches between the host and the storage solution support and are configured for the same MTU.

  3. Update Third-Party Appliance MTU:

    • Log into the management interface of the third-party storage solution and update the network settings to match the ESXi MTU (9000).

  4. Verify with vmkping:

    • Run the following command from the ESXi CLI to ensure large packets are passing without fragmentation: vmkping <vmkX> -d -s 8972 <Storage_IP>

  5. Retry Restore/vMotion:

    • Once the MTU is consistent, the APD condition should clear, allowing the migration to complete successfully.

Powered by Wolken Software