The system enters an state that is hard for the customer to get out of without a VM shutdown or config changes.

4.19.9

Always

1. Have some setting that makes the virt-launcher Pod get memory requests, i.e.: customer using higherWorkloadDensity/memoryOvercommitPercentage, but still allows for memory hotplug (don't set requests on the VM itself).

2. Create a ResourceQuota such as below, allowing up to 8Gi

apiVersion: v1
kind: ResourceQuota
metadata:
  name: vms
  namespace: openshift-mtv
spec:
  hard:
    requests.cpu: "2"
    requests.memory: 8Gi

2. Create a VM with 1 CPU and 4Gi of RAM (memory.guest = 4Gi)

3. Run the VM

4. Update the VM to 16G (which will request 100% / 150% * 16Gi = 10.6Gi) if overcommit is 150%.

5. Note 10.6Gi > quota of 8Gi

6. Now the controller keeps creating workload updates VMIM forever, and they are all being denied. Canceling migration doesn't work because the controller creates again.

kind: VirtualMachineInstanceMigration
...
status:
  conditions:
  - lastProbeTime: "2025-10-01T23:18:59Z"
    lastTransitionTime: null
    status: "True"
    type: migrationRejectedByResourceQuota


7. Only way out is to shutdown the VM or disable workload updates. This is undesirable for the user.

Keeps trying to migrate and generating tons of errors. Hard to get out of this situation.

7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-zfnfw" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-6lv9j" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-5sjrv" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-w8nv7" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-wklbk" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-rvjs9" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-l9pgn" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
7s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-cfdkf" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi
6s          Warning   FailedCreate           virtualmachineinstance/windows-2019                              Error creating pod: pods "virt-launcher-windows-2019-l5z46" is forbidden: exceeded quota: vms, requested: requests.memory=17521053377, used: requests.memory=4501933760, limited: requests.memory=8Gi

- Something more user friendly. 
- Indicate to the user workload update failed with clear reason (in the Console)
- Maybe abort once with a proper error and indicate to the user live migration will not be possible due to quota exceeded
- Perhaps suggest to shutdown and start, or disable workload update for this VM with clear indication and let user handle it, but stop this loop.