[ Tzahi 30-10-25 ]
after i have synced with bodnopoz@redhat.com 
i rename the bug ticket and added more info : 

Currently, it seems that for copy-offload migrations, the cost function is not correctly calculated.
As a result, several of populte pods migrations start on the same ESXi host simultaneously, despite the configuration limiting active migrations to 2.
This causes parallel migrations to fail, effectively ignoring the "max_vm_inflight"  setting.

Additionally, when using vmkfstools, VMware enforces a limitation of 2 active migrations in total.
Because of this, starting more migrations than allowed can crash the plan.

In the ECOSYSTEM environment, it sometimes manages to exceed the 2 migration limit per ESXi host without causing identical errors or failures, but this behavior remains inconsistent.

Storage-Offload Test Case 3.6 – 50 VMs, 2 disks, 1 ESXi host

The test failed during the initialization step on all 50 VMs with the following error:

DiskAllocationCompleted 
tasksDanger alert:Error
                       detailspopulator pod failed for PVC 20cb87b8-4e68-4553-a655-748aa34ffebb. Please check the pod log