During the cross-site cold migration, the process fails and returns the following errors in the VM-Monitor-Tasks:
"A fatal internal error occurred. See the virtual machine logs for more details. The migration transfer failed during the receive operation to socket 73801495571968: received 0/36 bytes. Connection reset by peer."
"VMotionstream timed out while waiting for disk 0's queue count to drop below the maximum limit of 32768 blocks. This could indicate either network or storage problems preventing proper block transfer. Failed to copy source vmdk. Timeout. Failed to copy one or more disks."
vCenter 8.0U3
Although the migration is intended to be a "cold" (powered-off) transfer, the operation fails because it times out on the vMotion data stream due to network latency and bandwidth constraints across the remote WAN link. The underlying trigger for this behavior is that the virtual machine has Per-VM Enhanced vMotion Compatibility (EVC) enabled in its configuration. In certain cross-vCenter or long-distance topologies, leaving Per-VM EVC enabled inadvertently forces the migration task to utilize highly sensitive vMotion stream protocols (to strictly validate the CPU baseline) rather than the standard, more resilient Network File Copy (NFC) protocol typically used for cold transfers. Because the vMotion stream is utilized over a remote WAN link, the memory buffers quickly saturate (hitting the 32,768 block limit) and fatally abort the connection.
To resolve this issue, manually disable Per-VM EVC on the virtual machine to allow the cold migration to proceed normally over the WAN link.
In the vSphere Client, select the affected virtual machine.
Navigate to Configure > More (or Settings) > VMware EVC.
If it is enabled, click Edit.
Select Disable EVC completely and click OK.
Attempt the cross-site cold migration again.
(Note: Once the VM successfully migrates and rests on the production hosts, you can re-enable Per-VM EVC if your architectural baseline requires it.)