Live migration of VMs take multiple hours to complete

Issue

• VMs running memory intensive workloads, such as RHOCP worker nodes, take multiple hours to complete a live migration, or times out a live migration.
• During live migration, openstack server show -c progress <server id> remains with the same value for hours. even though memory is still being copied from source to target node

• nova-compute.log repeatedly shows the following messages:

  2026-01-01 00:00:00.000 2 INFO nova.virt.libvirt.driver [req-aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa - - - - -] [instance: aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa] Migration running for 244 secs, memory 0% remaining (bytes processed=4275183238, remaining=19784058, total=4294967296); disk 100% remaining (bytes processed=0, remaining=0, total=0).