This issue has been resolved.

An unintended network restart on some DGP hypervisors disrupted the block storage connection, causing several virtual machines to freeze.

We initially migrated affected VMs, but they remained frozen. After manually unfreezing the instances, all services recovered.

A subnet of the Virtual Machine Instances and Kubernetes Machine instnaces have been affected. We are working on restoring the services.

Some VMs and compute workloads running on dgp instance types have some network issues. We are working to investigate the issues.