allowAutoConverge[0] can be enabled via a MigrationPolicy or the HCO in the CLI or the UI.

It is possible[1] the throttling of the VM CPU may slow down the dirty rate during migration, allowing for the migration to progress faster and complete before the timeout.

However, this feature may disrupt the VM workload and make it non-responsive. In a busy cluster with high load, it is possible this feature is desirable.

How does this feature affect VM workloads on a cluster under-load at scale?

[0] https://issues.redhat.com/browse/CNV-48753

[1] https://github.com/kubevirt/kubevirt/issues/3504