Cis Ocp Vm V100 2 2

CIS 2.2 — Disable overcommitting guest memory

Profile Applicability

• Level 1

Description

The overcommitGuestOverhead configuration option enables the request for additional virtual machine management memory inside the virt-launcher pod. The overcommit feature is used to increase virtual machine density on the node, as long as the virtual machine doesn't request all the memory that it would need if fully loaded. However, if the VM were to use all of the memory it could, this would lead to the OpenShift Scheduler killing the workload.

Rationale

This is a hypervisor-level feature that allows nodes to host more virtual machines than would normally be allowed by KubeVirt's scheduling algorithms. If the VM consumes the entire memory might cause the guest to crash with workload interruptions and guest malfunctioning.

Impact

overCommitGuestOverhead is used to (unsafely)increase workload density, meaning more virtual machines might be able to run on a node than otherwise.

Audit Procedure

Verify that no virtual machines are using the overcommitGuestOverhead setting:

$ oc get vm -ojson -A |jq '.items[] | select(.spec.template.spec.domain.resources.overcommitGuestOverhead == true)| .metadata.namespace + "/" + .metadata.name'

Remediation

Stop any guest with this flag enabled and remove the flag before restarting. This flag can be removed on a given VM with the following command:

$ oc patch vm <vm-name> --type='json' -p='[
    {"op": "remove", "path": "/spec/template/spec/domain/resources/overcommitGuestOverhead"},
]'

Default Value

OpenShift Virtualization disables overcommitGuestOverhead by default and the list of virtual machines returned by the command should be empty.

References

None listed in the PDF.

CIS Controls

MITRE ATT&CK Mappings

Profile

• Level 1 - OpenShift Virtualization