Files
cozystack/packages/system/kubevirt/values.yaml
T
Aleksei Sviridkin 8dbc7bceb5 feat(kubevirt)!: drop HotplugVolumes for declarative volume hotplug
KubeVirt 1.9 deprecates HotplugVolumes in favour of
DeclarativeHotplugVolumes, which is Beta and therefore on by default,
but virt-controller skips declarative reconciliation for as long as the
old gate is on. Keeping it was only needed because kubevirt-csi
detached through the VMI-level removevolume, which virt-api accepts on
a VM-owned VMI only with the old gate. kubevirt-csi can now skip that
call, so its fallback goes off together with the gate. The two live in
different HelmReleases and roll out one after the other; in between the
driver may issue a VMI-level call virt-api rejects, and the unpublish
fails and is retried until both sides agree.

The chart now renders --enable-vmi-hotplug-fallback=false, which a
kubevirt-csi image built before the flag existed rejects at startup.
The driver image is stamped at release preparation, so this needs the
rebuilt image that ships in the same release.

DeclarativeHotplugVolumes is now listed explicitly. With HotplugVolumes
gone it is the only hotplug path, and disabling it through
disabledFeatureGates would make virt-api refuse every addvolume, so
the chart refuses that render instead.

This changes volume hotplug platform-wide. A VM-level addvolume or
removevolume now edits VM.spec directly and virt-controller reconciles
the VMI hotplug volumes against it, instead of going through
VM.status.volumeRequests. On the first reconcile it also removes every
hotpluggable VMI volume missing from VM.spec; platform migration 59
copies those into VM.spec beforehand on the variants that run
migrations.

Assisted-by: LLM
Signed-off-by: Aleksei Sviridkin <f@lex.la>
2026-09-28 23:37:48 +03:00

80 lines
3.6 KiB
YAML

# Injected by the platform from the cozystack-values Secret. The empty
# default keeps the chart renderable on its own, outside a cluster.
_cluster: {}
# CPU oversubscription ratio. When set, propagates to
# spec.configuration.developerConfiguration.cpuAllocationRatio on the
# KubeVirt CR. Driven by .Values.resources.cpuAllocationRatio at the
# platform level.
# cpuAllocationRatio: 10
# Feature gates appended after the static defaults
# (DeclarativeHotplugVolumes/ExpandDisks/LiveMigration/AutoResourceLimitsGate/CPUManager/GPU/VMExport).
# Empty by default. Filled by the platform with "HostDevices" when the
# gpu-operator package is enabled with the passthrough variant — the
# admission webhook rejects spec.template.spec.domain.devices.hostDevices
# without that gate even with the GPU gate on, since current KubeVirt
# splits them.
extraFeatureGates: []
# Feature gates appended after the static opt-outs under
# spec.configuration.developerConfiguration.disabledFeatureGates. The chart
# always disables Template and ExternalNetResourceInjection, and adds
# ImageVolume below Kubernetes 1.35, where the apiserver's own gate is off and
# would strip the launcher's image volumes. KubeVirt enables every Beta gate by
# default; list a gate here to switch off one this cluster cannot carry, e.g. on
# a container runtime without image volume support at 1.35 or newer.
# Naming a gate the chart already disables is a no-op: it renders once.
#
# Admins set this through spec.components.platform.values.kubevirt on the
# cozystack.cozystack-platform Package; the iaas bundle forwards it into the
# emitted cozystack.kubevirt Package as components.kubevirt.values.
#
# A gate this chart already enables cannot be switched off here, and a gate it
# disables cannot be switched back on through extraFeatureGates: KubeVirt
# refuses a CR that names the same gate in both lists, so the template stops
# the render and names the gate. ImageVolume is therefore rejected in
# extraFeatureGates below 1.35 and accepted from 1.35 on.
disabledFeatureGates: []
# Permitted host devices the KubeVirt admission webhook accepts under
# spec.template.spec.domain.devices.hostDevices / gpus on a VirtualMachine.
# Empty by default — leaves the CR's permittedHostDevices unset.
#
# Filled by the platform with NVIDIA PCI vendor:device → resourceName
# pairs (vendor 10DE) when gpu-operator runs in vfio passthrough mode,
# and with mediatedDevices entries when it runs in vGPU mode.
# externalResourceProvider: true is required because the resource names
# are published by gpu-operator's nvidia-sandbox-device-plugin DaemonSet,
# not by KubeVirt itself.
#
# Shape (subset of kubevirt.io/v1 PermittedHostDevices):
# pciHostDevices:
# - pciVendorSelector: "10DE:2335"
# resourceName: nvidia.com/GH100_H200_SXM_141GB
# externalResourceProvider: true
# mediatedDevices:
# - mdevNameSelector: NVIDIA L40S-24Q
# resourceName: nvidia.com/L40S-24Q
permittedHostDevices: {}
# mediatedDevicesConfiguration for vGPU. Filled by the platform when the
# gpu-operator package runs in vGPU mode; empty otherwise.
#
# Shape (subset of kubevirt.io/v1 MediatedDevicesConfiguration):
# mediatedDeviceTypes:
# - nvidia-740 # L40S-24Q
# nodeMediatedDeviceTypes:
# - nodeSelector:
# nvidia.com/gpu.product: NVIDIA-L40S
# mediatedDeviceTypes:
# - nvidia-740
mediatedDevicesConfiguration: {}
# Passed to spec.configuration.migrations on the KubeVirt CR.
# Empty leaves KubeVirt's migration defaults unchanged.
# bandwidthPerMigration is a quantity in bytes per second.
# Set through spec.components.platform.values.kubevirt.migrations
# on the cozystack.cozystack-platform Package.
migrations: {}