When an Azure Linux monitoring server becomes inaccessible due to SSH timeouts, corrupted VM agents, and a failed OS disk swap, standard repair workflows often fail. This real-world war story details a critical Azure infrastructure issue where the VM's guest OS was intact, but the control plane metadata—VM identity, agent channels, and extension registry—was poisoned. The solution involves a 'lift-and-shift' migration rather than traditional repair.

The process includes deallocating the broken VM, detaching the healthy OS disk, and attaching it to a newly created VM instance. This bypasses corrupted metadata, effectively transplanting the system into a fresh identity. This method is superior to disk repair for resolving identity corruption and proves that the VM object, not the disk, can be the root cause of persistent outages.