KSM Management

Kernel Samepage Merging (KSM) allows de-duplication of memory. KSM tries to find identical Memory Pages and merge those to free memory.

Further Information:
- KSM (Kernel Samepage Merging) feature_feature)
- Kernel Same-page Merging (KSM)

Enabling KSM through KubeVirt CR

KSM can be enabled on nodes by spec.configuration.ksmConfiguration in the KubeVirt CR.
ksmConfiguration instructs on which nodes KSM will be enabled, exposing a nodeLabelSelector.
nodeLabelSelector is a LabelSelector and defines the filter, based on the node labels. If a node’s labels match the label selector term, then on that node, KSM will be enabled.

NOTE
If nodeLabelSelector is nil KSM will not be enabled on any nodes.
Empty nodeLabelSelector will enable KSM on every node.

Examples:

  • Enabling KSM on nodes in which the hostname is node01 or node03:

    1. spec:
    2.   configuration:
    3.     ksmConfiguration:
    4.       nodeLabelSelector:
    5.         matchExpressions:
    6.           - key: kubernetes.io/hostname
    7.             operator: In
    8.             values:
    9.               - node01
    10.               - node03

  • Enabling KSM on nodes with labels kubevirt.io/first-label: true, kubevirt.io/second-label: true:

    1. spec:
    2.   configuration:
    3.     ksmConfiguration:
    4.       nodeLabelSelector:
    5.         matchLabels:
    6.           kubevirt.io/first-label: "true"
    7.           kubevirt.io/second-label: "true"

  • Enabling KSM on every node:

    1. spec:
    2.   configuration:
    3.     ksmConfiguration:
    4.       nodeLabelSelector: {}

Annotation and restore mechanism

On those nodes where KubeVirt enables the KSM via configuration, an annotation will be added (kubevirt.io/ksm-handler-managed).
This annotation is an internal record to keep track of which nodes are currently managed by virt-handler, so that it is possible to distinguish which nodes should be restored in case of future ksmConfiguration changes.

Let’s imagine this scenario:

  1. There are 3 nodes in the cluster and one of them(node01) has KSM externally enabled.
  2. An admin patches the KubeVirt CR adding a ksmConfiguration which enables ksm for node02 and node03.
  3. After a while, an admin patches again the KubeVirt CR deleting the ksmConfiguration.

Thanks to the annotation, the virt-handler is able to disable ksm on only those nodes where it itself had enabled it(node02 node03), leaving the others unchanged (node01).

Node labelling

KubeVirt can discover on which nodes KSM is enabled and will mark them with a special label (kubevirt.io/ksm-enabled) with value true. This label can be used to schedule the vms in nodes with KSM enabled or not.

  1. apiVersion: kubevirt.io/v1
  2. kind: VirtualMachine
  3. metadata:
  4.   name: testvm
  5. spec:
  6.   runStrategy: Always
  7.   template:
  8.     metadata:
  9.       labels:
  10.         kubevirt.io/vm: testvm
  11.     spec:
  12.       nodeSelector:
  13.         kubevirt.io/ksm-enabled: "true"
  14.       [...]