feat(hami): add HAMi GPU virtualization system package (#2484)
## What this PR does Documentation: https://github.com/cozystack/website/pull/517 Integrates [HAMi](https://github.com/Project-HAMi/HAMi) v2.8.1 (CNCF Sandbox) into Cozystack as a system-level package for fractional GPU sharing in tenant Kubernetes clusters. The integration covers three layers: - **System chart** (`packages/system/hami/`): Vendors upstream HAMi Helm chart with device plugin, scheduler extender, mutating webhook, and RuntimeClass configuration. The broken DRA subchart was removed — it renders resources even when disabled and references unpublished images. - **Kubernetes app addon** (`packages/apps/kubernetes/`): HAMi exposed as an optional toggle (`hami.enabled`). Automatically disables GPU Operator's native device plugin when active to avoid conflicts. Enforces hard dependency on GPU Operator. - **Platform registration** (`packages/core/platform/`): HAMi declared as PackageSource with gpu-operator dependency, included in the iaas bundle. **Known limitation**: HAMi-core relies on a private glibc symbol (`_dl_sym`) removed in glibc 2.34, which breaks compute isolation on modern container images (Ubuntu 22.04+, PyTorch/TensorFlow official images). Alpine/musl is entirely incompatible. See the package README for details and upstream issue references. ### Screenshots N/A — no UI changes. ### Release note ```release-note feat(hami): add HAMi GPU virtualization as an optional system package for fractional GPU sharing. Enables memory and compute isolation for NVIDIA GPUs across tenant workloads. Integrates with GPU Operator and can be enabled per-cluster via the hami.enabled toggle. Note: compute isolation requires glibc < 2.34 in workload containers. ``` <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **New Features** * Added HAMi GPU virtualization middleware as an optional cluster addon to enable fractional GPU sharing. * HAMi addon includes an enable toggle (default: disabled) and a customizable Helm values override; when enabled it enforces GPU Operator presence and adjusts GPU Operator values accordingly. * New HelmRelease integration for deploying HAMi with dependency ordering and conditional rendering. * **Documentation** * Added HAMi setup guide, compatibility notes, and updated cluster addon configuration reference. <!-- end of auto-generated comment: release notes by coderabbit.ai -->
This commit is contained in:
Aleksei Sviridkin
GitHub
commit
f45facca77
50 changed files with 3281 additions and 4 deletions
|
|
@ -69,6 +69,9 @@ type Addons struct {
|
|||
// NVIDIA GPU Operator.
|
||||
// +kubebuilder:default:={}
|
||||
GpuOperator GPUOperatorAddon `json:"gpuOperator"`
|
||||
// HAMi GPU virtualization middleware.
|
||||
// +kubebuilder:default:={}
|
||||
Hami HAMiAddon `json:"hami"`
|
||||
// Ingress-NGINX controller.
|
||||
// +kubebuilder:default:={}
|
||||
IngressNginx IngressNginxAddon `json:"ingressNginx"`
|
||||
|
|
@ -160,6 +163,15 @@ type GatewayAPIAddon struct {
|
|||
Enabled bool `json:"enabled"`
|
||||
}
|
||||
|
||||
type HAMiAddon struct {
|
||||
// Enable HAMi (requires GPU Operator).
|
||||
// +kubebuilder:default:=false
|
||||
Enabled bool `json:"enabled"`
|
||||
// Custom Helm values overrides.
|
||||
// +kubebuilder:default:={}
|
||||
ValuesOverride k8sRuntime.RawExtension `json:"valuesOverride"`
|
||||
}
|
||||
|
||||
type Images struct {
|
||||
// Image used by the wait-for-kubeconfig init container. Empty falls back to images/busybox.tag.
|
||||
// +kubebuilder:default:=""
|
||||
|
|
|
|||
|
|
@ -49,6 +49,7 @@ func (in *Addons) DeepCopyInto(out *Addons) {
|
|||
in.Fluxcd.DeepCopyInto(&out.Fluxcd)
|
||||
out.GatewayAPI = in.GatewayAPI
|
||||
in.GpuOperator.DeepCopyInto(&out.GpuOperator)
|
||||
in.Hami.DeepCopyInto(&out.Hami)
|
||||
in.IngressNginx.DeepCopyInto(&out.IngressNginx)
|
||||
in.MonitoringAgents.DeepCopyInto(&out.MonitoringAgents)
|
||||
in.Velero.DeepCopyInto(&out.Velero)
|
||||
|
|
@ -135,6 +136,7 @@ func (in *ConfigSpec) DeepCopyInto(out *ConfigSpec) {
|
|||
}
|
||||
in.Addons.DeepCopyInto(&out.Addons)
|
||||
in.ControlPlane.DeepCopyInto(&out.ControlPlane)
|
||||
out.Images = in.Images
|
||||
}
|
||||
|
||||
// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new ConfigSpec.
|
||||
|
|
@ -260,6 +262,37 @@ func (in *GatewayAPIAddon) DeepCopy() *GatewayAPIAddon {
|
|||
return out
|
||||
}
|
||||
|
||||
// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
|
||||
func (in *HAMiAddon) DeepCopyInto(out *HAMiAddon) {
|
||||
*out = *in
|
||||
in.ValuesOverride.DeepCopyInto(&out.ValuesOverride)
|
||||
}
|
||||
|
||||
// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new HAMiAddon.
|
||||
func (in *HAMiAddon) DeepCopy() *HAMiAddon {
|
||||
if in == nil {
|
||||
return nil
|
||||
}
|
||||
out := new(HAMiAddon)
|
||||
in.DeepCopyInto(out)
|
||||
return out
|
||||
}
|
||||
|
||||
// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
|
||||
func (in *Images) DeepCopyInto(out *Images) {
|
||||
*out = *in
|
||||
}
|
||||
|
||||
// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new Images.
|
||||
func (in *Images) DeepCopy() *Images {
|
||||
if in == nil {
|
||||
return nil
|
||||
}
|
||||
out := new(Images)
|
||||
in.DeepCopyInto(out)
|
||||
return out
|
||||
}
|
||||
|
||||
// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
|
||||
func (in *IngressNginxAddon) DeepCopyInto(out *IngressNginxAddon) {
|
||||
*out = *in
|
||||
|
|
|
|||
|
|
@ -128,6 +128,9 @@ See the reference for components utilized in this service:
|
|||
| `addons.gpuOperator` | NVIDIA GPU Operator. | `object` | `{}` |
|
||||
| `addons.gpuOperator.enabled` | Enable GPU Operator. | `bool` | `false` |
|
||||
| `addons.gpuOperator.valuesOverride` | Custom Helm values overrides. | `object` | `{}` |
|
||||
| `addons.hami` | HAMi GPU virtualization middleware. | `object` | `{}` |
|
||||
| `addons.hami.enabled` | Enable HAMi (requires GPU Operator). | `bool` | `false` |
|
||||
| `addons.hami.valuesOverride` | Custom Helm values overrides. | `object` | `{}` |
|
||||
| `addons.fluxcd` | FluxCD GitOps operator. | `object` | `{}` |
|
||||
| `addons.fluxcd.enabled` | Enable FluxCD. | `bool` | `false` |
|
||||
| `addons.fluxcd.valuesOverride` | Custom Helm values overrides. | `object` | `{}` |
|
||||
|
|
|
|||
|
|
@ -1,3 +1,11 @@
|
|||
{{- define "cozystack.defaultGpuOperatorValues" -}}
|
||||
{{- if .Values.addons.hami.enabled }}
|
||||
gpu-operator:
|
||||
devicePlugin:
|
||||
enabled: false
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
|
||||
{{- if and .Values.addons.gpuOperator.enabled .Values._namespace.etcd }}
|
||||
apiVersion: helm.toolkit.fluxcd.io/v2
|
||||
kind: HelmRelease
|
||||
|
|
@ -29,9 +37,12 @@ spec:
|
|||
force: true
|
||||
remediation:
|
||||
retries: -1
|
||||
{{- with .Values.addons.gpuOperator.valuesOverride }}
|
||||
{{- $defaults := fromYaml (include "cozystack.defaultGpuOperatorValues" .) }}
|
||||
{{- $overrides := deepCopy (default (dict) .Values.addons.gpuOperator.valuesOverride) }}
|
||||
{{- $merged := mergeOverwrite (default (dict) $defaults) $overrides }}
|
||||
{{- if $merged }}
|
||||
values:
|
||||
{{- toYaml . | nindent 4 }}
|
||||
{{- toYaml $merged | nindent 4 }}
|
||||
{{- end }}
|
||||
|
||||
dependsOn:
|
||||
|
|
|
|||
49
packages/apps/kubernetes/templates/helmreleases/hami.yaml
Normal file
49
packages/apps/kubernetes/templates/helmreleases/hami.yaml
Normal file
|
|
@ -0,0 +1,49 @@
|
|||
{{- if and .Values.addons.hami.enabled .Values._namespace.etcd }}
|
||||
{{- if not .Values.addons.gpuOperator.enabled }}
|
||||
{{- fail "addons.hami requires addons.gpuOperator to be enabled" }}
|
||||
{{- end }}
|
||||
apiVersion: helm.toolkit.fluxcd.io/v2
|
||||
kind: HelmRelease
|
||||
metadata:
|
||||
name: {{ .Release.Name }}-hami
|
||||
labels:
|
||||
cozystack.io/repository: system
|
||||
cozystack.io/target-cluster-name: {{ .Release.Name }}
|
||||
sharding.fluxcd.io/key: tenants
|
||||
spec:
|
||||
releaseName: hami
|
||||
chartRef:
|
||||
kind: ExternalArtifact
|
||||
name: cozystack-kubernetes-application-kubevirt-kubernetes-hami
|
||||
namespace: cozy-system
|
||||
kubeConfig:
|
||||
secretRef:
|
||||
name: {{ .Release.Name }}-admin-kubeconfig
|
||||
key: super-admin.svc
|
||||
targetNamespace: cozy-hami
|
||||
storageNamespace: cozy-hami
|
||||
interval: 5m
|
||||
timeout: 10m
|
||||
install:
|
||||
createNamespace: true
|
||||
remediation:
|
||||
retries: -1
|
||||
upgrade:
|
||||
force: true
|
||||
remediation:
|
||||
retries: -1
|
||||
{{- with .Values.addons.hami.valuesOverride }}
|
||||
values:
|
||||
{{- toYaml . | nindent 4 }}
|
||||
{{- end }}
|
||||
|
||||
dependsOn:
|
||||
{{- if lookup "helm.toolkit.fluxcd.io/v2" "HelmRelease" .Release.Namespace .Release.Name }}
|
||||
- name: {{ .Release.Name }}
|
||||
namespace: {{ .Release.Namespace }}
|
||||
{{- end }}
|
||||
- name: {{ .Release.Name }}-cilium
|
||||
namespace: {{ .Release.Namespace }}
|
||||
- name: {{ .Release.Name }}-gpu-operator
|
||||
namespace: {{ .Release.Namespace }}
|
||||
{{- end }}
|
||||
99
packages/apps/kubernetes/tests/gpu_operator_hami_test.yaml
Normal file
99
packages/apps/kubernetes/tests/gpu_operator_hami_test.yaml
Normal file
|
|
@ -0,0 +1,99 @@
|
|||
suite: GPU Operator HelmRelease HAMi integration tests
|
||||
templates:
|
||||
- templates/helmreleases/gpu-operator.yaml
|
||||
values:
|
||||
- values-ci.yaml
|
||||
tests:
|
||||
- it: should disable devicePlugin when hami is enabled
|
||||
set:
|
||||
addons:
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- equal:
|
||||
path: spec.values.gpu-operator.devicePlugin.enabled
|
||||
value: false
|
||||
|
||||
- it: should not have values when hami is disabled and no overrides
|
||||
set:
|
||||
addons:
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
hami:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- notExists:
|
||||
path: spec.values
|
||||
|
||||
- it: should apply hami defaults when valuesOverride key is omitted
|
||||
set:
|
||||
addons:
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
hami:
|
||||
enabled: true
|
||||
asserts:
|
||||
- equal:
|
||||
path: spec.values.gpu-operator.devicePlugin.enabled
|
||||
value: false
|
||||
|
||||
- it: should allow user overrides to merge with hami defaults
|
||||
set:
|
||||
addons:
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride:
|
||||
gpu-operator:
|
||||
driver:
|
||||
enabled: false
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- equal:
|
||||
path: spec.values.gpu-operator.devicePlugin.enabled
|
||||
value: false
|
||||
- equal:
|
||||
path: spec.values.gpu-operator.driver.enabled
|
||||
value: false
|
||||
|
||||
- it: should let user explicitly override devicePlugin.enabled to true with hami enabled
|
||||
set:
|
||||
addons:
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride:
|
||||
gpu-operator:
|
||||
devicePlugin:
|
||||
enabled: true
|
||||
driver:
|
||||
enabled: false
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- equal:
|
||||
path: spec.values.gpu-operator.devicePlugin.enabled
|
||||
value: true
|
||||
- equal:
|
||||
path: spec.values.gpu-operator.driver.enabled
|
||||
value: false
|
||||
|
||||
- it: should not render when gpuOperator is disabled
|
||||
set:
|
||||
addons:
|
||||
gpuOperator:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
hami:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- hasDocuments:
|
||||
count: 0
|
||||
153
packages/apps/kubernetes/tests/hami_test.yaml
Normal file
153
packages/apps/kubernetes/tests/hami_test.yaml
Normal file
|
|
@ -0,0 +1,153 @@
|
|||
suite: HAMi HelmRelease tests
|
||||
templates:
|
||||
- templates/helmreleases/hami.yaml
|
||||
values:
|
||||
- values-ci.yaml
|
||||
tests:
|
||||
- it: should not render when hami is disabled
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- hasDocuments:
|
||||
count: 0
|
||||
|
||||
- it: should render HelmRelease when hami is enabled
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- hasDocuments:
|
||||
count: 1
|
||||
- isKind:
|
||||
of: HelmRelease
|
||||
|
||||
- it: should fail when gpuOperator is not enabled
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- failedTemplate:
|
||||
errorMessage: "addons.hami requires addons.gpuOperator to be enabled"
|
||||
|
||||
- it: should have correct metadata labels
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- equal:
|
||||
path: metadata.labels["cozystack.io/repository"]
|
||||
value: system
|
||||
- equal:
|
||||
path: metadata.labels["sharding.fluxcd.io/key"]
|
||||
value: tenants
|
||||
|
||||
- it: should use ExternalArtifact chartRef
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- equal:
|
||||
path: spec.chartRef.kind
|
||||
value: ExternalArtifact
|
||||
- equal:
|
||||
path: spec.chartRef.namespace
|
||||
value: cozy-system
|
||||
|
||||
- it: should target cozy-hami namespace
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- equal:
|
||||
path: spec.targetNamespace
|
||||
value: cozy-hami
|
||||
- equal:
|
||||
path: spec.storageNamespace
|
||||
value: cozy-hami
|
||||
|
||||
- it: should depend on gpu-operator and cilium
|
||||
release:
|
||||
name: test
|
||||
namespace: test-ns
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- contains:
|
||||
path: spec.dependsOn
|
||||
content:
|
||||
name: test-cilium
|
||||
namespace: test-ns
|
||||
- contains:
|
||||
path: spec.dependsOn
|
||||
content:
|
||||
name: test-gpu-operator
|
||||
namespace: test-ns
|
||||
|
||||
- it: should not render spec.values when valuesOverride is empty
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- hasDocuments:
|
||||
count: 1
|
||||
- notExists:
|
||||
path: spec.values
|
||||
|
||||
- it: should pass through valuesOverride
|
||||
set:
|
||||
addons:
|
||||
hami:
|
||||
enabled: true
|
||||
valuesOverride:
|
||||
hami:
|
||||
devicePlugin:
|
||||
deviceSplitCount: 5
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
valuesOverride: {}
|
||||
asserts:
|
||||
- equal:
|
||||
path: spec.values.hami.devicePlugin.deviceSplitCount
|
||||
value: 5
|
||||
|
|
@ -149,6 +149,7 @@
|
|||
"fluxcd",
|
||||
"gatewayAPI",
|
||||
"gpuOperator",
|
||||
"hami",
|
||||
"ingressNginx",
|
||||
"monitoringAgents",
|
||||
"velero",
|
||||
|
|
@ -268,6 +269,28 @@
|
|||
}
|
||||
}
|
||||
},
|
||||
"hami": {
|
||||
"description": "HAMi GPU virtualization middleware.",
|
||||
"type": "object",
|
||||
"default": {},
|
||||
"required": [
|
||||
"enabled",
|
||||
"valuesOverride"
|
||||
],
|
||||
"properties": {
|
||||
"enabled": {
|
||||
"description": "Enable HAMi (requires GPU Operator).",
|
||||
"type": "boolean",
|
||||
"default": false
|
||||
},
|
||||
"valuesOverride": {
|
||||
"description": "Custom Helm values overrides.",
|
||||
"type": "object",
|
||||
"default": {},
|
||||
"x-kubernetes-preserve-unknown-fields": true
|
||||
}
|
||||
}
|
||||
},
|
||||
"ingressNginx": {
|
||||
"description": "Ingress-NGINX controller.",
|
||||
"type": "object",
|
||||
|
|
|
|||
|
|
@ -94,6 +94,10 @@ host: ""
|
|||
## @field {bool} enabled - Enable FluxCD.
|
||||
## @field {object} valuesOverride - Custom Helm values overrides.
|
||||
|
||||
## @typedef {struct} HAMiAddon - HAMi GPU virtualization middleware.
|
||||
## @field {bool} enabled - Enable HAMi (requires GPU Operator).
|
||||
## @field {object} valuesOverride - Custom Helm values overrides.
|
||||
|
||||
## @typedef {struct} MonitoringAgentsAddon - Monitoring agents (Fluent Bit, VMAgents).
|
||||
## @field {bool} enabled - Enable monitoring agents.
|
||||
## @field {object} valuesOverride - Custom Helm values overrides.
|
||||
|
|
@ -114,6 +118,7 @@ host: ""
|
|||
## @field {GatewayAPIAddon} gatewayAPI - Gateway API addon.
|
||||
## @field {IngressNginxAddon} ingressNginx - Ingress-NGINX controller.
|
||||
## @field {GPUOperatorAddon} gpuOperator - NVIDIA GPU Operator.
|
||||
## @field {HAMiAddon} hami - HAMi GPU virtualization middleware.
|
||||
## @field {FluxCDAddon} fluxcd - FluxCD GitOps operator.
|
||||
## @field {MonitoringAgentsAddon} monitoringAgents - Monitoring agents.
|
||||
## @field {VerticalPodAutoscalerAddon} verticalPodAutoscaler - Vertical Pod Autoscaler.
|
||||
|
|
@ -137,6 +142,9 @@ addons:
|
|||
gpuOperator:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
hami:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
fluxcd:
|
||||
enabled: false
|
||||
valuesOverride: {}
|
||||
|
|
|
|||
24
packages/core/platform/sources/hami.yaml
Normal file
24
packages/core/platform/sources/hami.yaml
Normal file
|
|
@ -0,0 +1,24 @@
|
|||
---
|
||||
apiVersion: cozystack.io/v1alpha1
|
||||
kind: PackageSource
|
||||
metadata:
|
||||
name: cozystack.hami
|
||||
spec:
|
||||
sourceRef:
|
||||
kind: OCIRepository
|
||||
name: cozystack-packages
|
||||
namespace: cozy-system
|
||||
path: /
|
||||
variants:
|
||||
- name: default
|
||||
dependsOn:
|
||||
- cozystack.gpu-operator
|
||||
components:
|
||||
- name: hami
|
||||
path: system/hami
|
||||
valuesFiles:
|
||||
- values.yaml
|
||||
install:
|
||||
privileged: true
|
||||
namespace: cozy-hami
|
||||
releaseName: hami
|
||||
|
|
@ -52,6 +52,8 @@ spec:
|
|||
path: system/cilium
|
||||
- name: kubernetes-gpu-operator
|
||||
path: system/gpu-operator
|
||||
- name: kubernetes-hami
|
||||
path: system/hami
|
||||
- name: kubernetes-vertical-pod-autoscaler
|
||||
path: system/vertical-pod-autoscaler
|
||||
- name: kubernetes-prometheus-operator-crds
|
||||
|
|
|
|||
|
|
@ -10,6 +10,7 @@
|
|||
{{include "cozystack.platform.package.default" (list "cozystack.kubevirt-cdi" $) }}
|
||||
{{include "cozystack.platform.package.optional.default" (list "cozystack.vm-default-images" $) }}
|
||||
{{include "cozystack.platform.package.optional.default" (list "cozystack.gpu-operator" $) }}
|
||||
{{include "cozystack.platform.package.optional.default" (list "cozystack.hami" $) }}
|
||||
{{include "cozystack.platform.package.default" (list "cozystack.kamaji" $) }}
|
||||
{{include "cozystack.platform.package.default" (list "cozystack.capi-operator" $) }}
|
||||
{{include "cozystack.platform.package.default" (list "cozystack.capi-provider-bootstrap-kubeadm" $) }}
|
||||
|
|
|
|||
3
packages/system/hami/Chart.yaml
Normal file
3
packages/system/hami/Chart.yaml
Normal file
|
|
@ -0,0 +1,3 @@
|
|||
apiVersion: v2
|
||||
name: cozy-hami
|
||||
version: 0.0.0 # Placeholder, the actual version will be automatically set during the build process
|
||||
43
packages/system/hami/Makefile
Normal file
43
packages/system/hami/Makefile
Normal file
|
|
@ -0,0 +1,43 @@
|
|||
export NAME=hami
|
||||
export NAMESPACE=cozy-$(NAME)
|
||||
|
||||
include ../../../hack/common-envs.mk
|
||||
include ../../../hack/package.mk
|
||||
|
||||
# When bumping the HAMi version, run `make update` and then review
|
||||
# the resulting diff in `charts/hami/`. The recipe below reproduces the
|
||||
# top-level vendoring overrides automatically:
|
||||
#
|
||||
# 1. Removes the broken hami-dra subchart. Upstream's NVIDIA DRA driver
|
||||
# path requires kubelet DRA support that cozystack does not enable
|
||||
# and has no upstream fix tracked. See commit 3c5521e.
|
||||
# 2. Empties Chart.yaml dependencies and drops Chart.lock so Helm does
|
||||
# not try to re-pull hami-dra at build time. See commit 2734dc0.
|
||||
# 3. Strips dra/hami-dra/podSecurityPolicy blocks from the upstream
|
||||
# values.yaml since the corresponding code paths are gone. PSP is
|
||||
# removed from Kubernetes 1.25+ and is unused by cozystack.
|
||||
#
|
||||
# Template-level patches are NOT reproduced automatically:
|
||||
#
|
||||
# * Scheduler templates have `{{- if .Values.dra.enabled }}` blocks
|
||||
# that need to be removed because the dra value is gone (commit
|
||||
# 2734dc0 stripped them).
|
||||
# * device-plugin/monitorservice.yaml uses `indent` with leading
|
||||
# whitespace; it must be rewritten to `nindent` for the labels block
|
||||
# to render correctly when devicePlugin.service.labels is set
|
||||
# (commit 3685254).
|
||||
#
|
||||
# After `make update`, run `git diff -- charts/hami/templates/` and
|
||||
# replay these template patches against the new upstream version, then
|
||||
# verify with `helm unittest`. If upstream restructured the affected
|
||||
# files, the patches may need to be redesigned rather than reapplied.
|
||||
|
||||
update:
|
||||
rm -rf charts
|
||||
helm repo add hami-charts https://project-hami.github.io/HAMi/
|
||||
helm repo update hami-charts
|
||||
helm pull hami-charts/hami --untar --untardir charts
|
||||
rm -rf charts/hami/charts/hami-dra
|
||||
yq --inplace '.dependencies = []' charts/hami/Chart.yaml
|
||||
rm -f charts/hami/Chart.lock
|
||||
yq --inplace 'del(.dra) | del(.["hami-dra"]) | del(.podSecurityPolicy)' charts/hami/values.yaml
|
||||
82
packages/system/hami/README.md
Normal file
82
packages/system/hami/README.md
Normal file
|
|
@ -0,0 +1,82 @@
|
|||
# HAMi — GPU Virtualization Middleware
|
||||
|
||||
[HAMi](https://github.com/Project-HAMi/HAMi) (Heterogeneous AI Computing Virtualization Middleware) is a CNCF Sandbox project that enables fractional GPU sharing in Kubernetes. It allows workloads to request specific amounts of GPU memory and compute cores instead of claiming entire GPUs.
|
||||
|
||||
## Architecture
|
||||
|
||||
HAMi consists of four components:
|
||||
|
||||
- **MutatingWebhook** — intercepts pod creation, injects `schedulerName: hami-scheduler`
|
||||
- **Scheduler Extender** — extends kube-scheduler with GPU-aware Filter and Bind logic
|
||||
- **Device Plugin** (DaemonSet) — registers vGPU resources via the Kubernetes Device Plugin API
|
||||
- **HAMi-core** (`libvgpu.so`) — `LD_PRELOAD` library injected into workload containers, intercepts CUDA API calls to enforce memory and compute isolation
|
||||
|
||||
## Prerequisites
|
||||
|
||||
- GPU Operator must be enabled (`addons.gpuOperator.enabled: true`)
|
||||
- NVIDIA driver >= 440 on host nodes
|
||||
- nvidia-container-toolkit configured as the default container runtime
|
||||
- GPU nodes labeled with `gpu=on`
|
||||
|
||||
## Known Limitations
|
||||
|
||||
### glibc < 2.34 requirement for workload containers
|
||||
|
||||
HAMi-core uses `LD_PRELOAD` to intercept `dlsym()` for CUDA symbol resolution. The fallback code path relies on `_dl_sym`, a private glibc internal symbol that was removed in glibc 2.34 when libdl and libpthread were merged into libc.so.
|
||||
|
||||
**This limitation affects workload containers only**, not the host OS or HAMi's own components.
|
||||
|
||||
| Distribution | glibc | Result |
|
||||
| --------------- | ----- | -------------------------------------------- |
|
||||
| Ubuntu 18.04 | 2.27 | Full isolation (memory + compute) |
|
||||
| Ubuntu 20.04 | 2.31 | Full isolation (memory + compute) |
|
||||
| Ubuntu 22.04 | 2.35 | Memory isolation works, compute breaks |
|
||||
| Ubuntu 24.04 | 2.39 | Both memory and compute isolation break |
|
||||
| Alpine (musl) | N/A | Completely incompatible (`dlvsym` absent) |
|
||||
|
||||
Most modern ML/AI base images (CUDA 12.x, PyTorch 2.x, TensorFlow 2.x) use Ubuntu 22.04+ with glibc >= 2.35, which means compute isolation will not work with these images until the upstream fix is merged.
|
||||
|
||||
**Upstream tracking issues:**
|
||||
|
||||
- [HAMi-core#174](https://github.com/Project-HAMi/HAMi-core/issues/174) — `_dl_sym` removal in glibc 2.34 breaks HAMi-core's CUDA symbol resolution at the symbol level
|
||||
- [HAMi#1190](https://github.com/Project-HAMi/HAMi/issues/1190) — maintainer thread confirming the empirical per-glibc-version isolation behavior shown in the table above
|
||||
|
||||
### musl libc (Alpine) incompatibility
|
||||
|
||||
HAMi-core is completely incompatible with musl libc. The `dlvsym()` function used by HAMi-core is a glibc extension not available in musl. Only glibc-based container images (Debian, Ubuntu, RHEL, etc.) can use HAMi GPU isolation.
|
||||
|
||||
## Usage
|
||||
|
||||
Enable HAMi in your tenant Kubernetes cluster values:
|
||||
|
||||
```yaml
|
||||
addons:
|
||||
gpuOperator:
|
||||
enabled: true
|
||||
hami:
|
||||
enabled: true
|
||||
```
|
||||
|
||||
When HAMi is enabled, GPU Operator's built-in device plugin is automatically disabled to avoid conflicts. This default is preserved by setting `addons.gpuOperator.valuesOverride.gpu-operator.devicePlugin.enabled: false`; advanced topologies that partition GPU pools (e.g. some nodes use HAMi while others run the standard NVIDIA device plugin via node selectors) can re-enable it explicitly through `valuesOverride`.
|
||||
|
||||
### Requesting fractional GPU resources
|
||||
|
||||
```yaml
|
||||
resources:
|
||||
limits:
|
||||
nvidia.com/gpu: 1
|
||||
nvidia.com/gpumem: 3000 # 3000 MB of GPU memory
|
||||
nvidia.com/gpucores: 30 # 30% of GPU compute cores
|
||||
```
|
||||
|
||||
## Parameters
|
||||
|
||||
Default values shown below are inherited from the upstream HAMi chart and may change with upstream updates.
|
||||
|
||||
| Name | Description | Default |
|
||||
| --- | --- | --- |
|
||||
| `hami.devicePlugin.runtimeClassName` | RuntimeClass for device plugin pods | `nvidia` |
|
||||
| `hami.devicePlugin.deviceSplitCount` | Max virtual GPUs per physical GPU | `10` |
|
||||
| `hami.devicePlugin.deviceMemoryScaling` | Memory overcommit factor (> 1.0 enables overcommit) | `1` |
|
||||
| `hami.scheduler.defaultSchedulerPolicy.nodeSchedulerPolicy` | Node packing strategy (`binpack` or `spread`) | `binpack` |
|
||||
| `hami.scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy` | GPU packing strategy (`binpack` or `spread`) | `spread` |
|
||||
18
packages/system/hami/charts/hami/Chart.yaml
Normal file
18
packages/system/hami/charts/hami/Chart.yaml
Normal file
|
|
@ -0,0 +1,18 @@
|
|||
apiVersion: v2
|
||||
appVersion: 2.8.1
|
||||
dependencies: []
|
||||
description: Heterogeneous AI Computing Virtualization Middleware
|
||||
keywords:
|
||||
- vgpu
|
||||
- gpu
|
||||
kubeVersion: '>= 1.18.0-0'
|
||||
maintainers:
|
||||
- email: archlitchi@gmail.com
|
||||
name: limengxuan
|
||||
- email: xiaozhang0210@hotmail.com
|
||||
name: zhangxiao
|
||||
name: hami
|
||||
sources:
|
||||
- https://github.com/Project-HAMi/HAMi
|
||||
type: application
|
||||
version: 2.8.1
|
||||
237
packages/system/hami/charts/hami/README.md
Normal file
237
packages/system/hami/charts/hami/README.md
Normal file
|
|
@ -0,0 +1,237 @@
|
|||
# HAMi Helm Chart Values Documentation
|
||||
|
||||
This document provides detailed descriptions of all configurable values parameters for the HAMi Helm Chart.
|
||||
|
||||
## Global Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `global.imageRegistry` | Global Docker image registry | `""` |
|
||||
| `global.imagePullSecrets` | Global Docker image pull secrets | `[]` |
|
||||
| `global.imageTag` | Image tag | `"v2.8.1"` |
|
||||
| `global.gpuHookPath` | GPU Hook path | `/usr/local` |
|
||||
| `global.labels` | Global labels | `{}` |
|
||||
| `global.annotations` | Global annotations | `{}` |
|
||||
| `global.managedNodeSelectorEnable` | Whether to enable managed node selector | `false` |
|
||||
| `global.managedNodeSelector.usage` | Managed node selector usage | `"gpu"` |
|
||||
| `nameOverride` | Name override | `""` |
|
||||
| `fullnameOverride` | Full name override | `""` |
|
||||
| `namespaceOverride` | Namespace override | `""` |
|
||||
|
||||
## Resource Name Configuration
|
||||
|
||||
### NVIDIA GPU Resources
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `resourceName` | GPU resource name | `"nvidia.com/gpu"` |
|
||||
| `resourceMem` | GPU memory resource name | `"nvidia.com/gpumem"` |
|
||||
| `resourceMemPercentage` | GPU memory percentage resource name | `"nvidia.com/gpumem-percentage"` |
|
||||
| `resourceCores` | GPU core resource name | `"nvidia.com/gpucores"` |
|
||||
| `resourcePriority` | GPU priority resource name | `"nvidia.com/priority"` |
|
||||
|
||||
### Cambricon MLU Resources
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `mluResourceName` | MLU resource name | `"cambricon.com/vmlu"` |
|
||||
| `mluResourceMem` | MLU memory resource name | `"cambricon.com/mlu.smlu.vmemory"` |
|
||||
| `mluResourceCores` | MLU core resource name | `"cambricon.com/mlu.smlu.vcore"` |
|
||||
|
||||
### Hygon DCU Resources
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `dcuResourceName` | DCU resource name | `"hygon.com/dcunum"` |
|
||||
| `dcuResourceMem` | DCU memory resource name | `"hygon.com/dcumem"` |
|
||||
| `dcuResourceCores` | DCU core resource name | `"hygon.com/dcucores"` |
|
||||
|
||||
### Metax GPU Resources
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `metaxResourceName` | GPU resource name | `"metax-tech.com/sgpu"` |
|
||||
| `metaxResourceCore` | GPU core resource name | `"metax-tech.com/vcore"` |
|
||||
| `metaxResourceMem` | GPU memory resource name | `"metax-tech.com/vmemory"` |
|
||||
| `metaxsGPUTopologyAware` | GPU topology awareness | `"false"` |
|
||||
|
||||
### Enflame GCU Resources
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `enflameResourceNameVGCU` | vGCU resource name | `"enflame.com/vgcu"` |
|
||||
| `enflameResourceNameVGCUPercentage` | vGCU percentage resource name | `"enflame.com/vgcu-percentage"` |
|
||||
|
||||
### Kunlunxin XPU Resources
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `kunlunResourceName` | XPU resource name | `"kunlunxin.com/xpu"` |
|
||||
|
||||
## Scheduler Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `schedulerName` | Scheduler name | `"hami-scheduler"` |
|
||||
| `scheduler.nodeName` | Define node name, scheduler will schedule to this node | `""` |
|
||||
| `scheduler.overwriteEnv` | Whether to overwrite environment variables | `"false"` |
|
||||
| `scheduler.defaultSchedulerPolicy.nodeSchedulerPolicy` | Node scheduler policy | `binpack` |
|
||||
| `scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy` | GPU scheduler policy | `spread` |
|
||||
| `scheduler.metricsBindAddress` | Metrics bind address | `":9395"` |
|
||||
| `scheduler.forceOverwriteDefaultScheduler` | Whether to force overwrite default scheduler | `true` |
|
||||
| `scheduler.livenessProbe` | Whether to enable liveness probe | `false` |
|
||||
| `scheduler.leaderElect` | Whether to enable leader election | `true` |
|
||||
| `scheduler.replicas` | Number of replicas | `1` |
|
||||
|
||||
### Kube Scheduler Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `scheduler.kubeScheduler.enabled` | Whether to run kube-scheduler container in scheduler pod | `true` |
|
||||
| `scheduler.kubeScheduler.image.registry` | Kube scheduler image registry | `"registry.cn-hangzhou.aliyuncs.com"` |
|
||||
| `scheduler.kubeScheduler.image.repository` | Kube scheduler image repository | `"google_containers/kube-scheduler"` |
|
||||
| `scheduler.kubeScheduler.image.tag` | Kube scheduler image tag | `""` |
|
||||
| `scheduler.kubeScheduler.image.pullPolicy` | Kube scheduler image pull policy | `IfNotPresent` |
|
||||
| `scheduler.kubeScheduler.image.pullSecrets` | Kube scheduler image pull secrets | `[]` |
|
||||
| `scheduler.kubeScheduler.extraNewArgs` | Extra new arguments | `["--config=/config/config.yaml", "-v=4"]` |
|
||||
| `scheduler.kubeScheduler.extraArgs` | Extra arguments | `["--policy-config-file=/config/config.json", "-v=4"]` |
|
||||
|
||||
### Extender Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `scheduler.extender.image.registry` | Scheduler extender image registry | `"docker.io"` |
|
||||
| `scheduler.extender.image.repository` | Scheduler extender image repository | `"projecthami/hami"` |
|
||||
| `scheduler.extender.image.tag` | Scheduler extender image tag | `""` |
|
||||
| `scheduler.extender.image.pullPolicy` | Scheduler extender image pull policy | `IfNotPresent` |
|
||||
| `scheduler.extender.image.pullSecrets` | Scheduler extender image pull secrets | `[]` |
|
||||
| `scheduler.extender.extraArgs` | Scheduler extender extra arguments | `["--debug", "-v=4"]` |
|
||||
|
||||
### Admission Webhook Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `scheduler.admissionWebhook.enabled` | Whether to enable admission webhook | `true` |
|
||||
| `scheduler.admissionWebhook.customURL.enabled` | Whether to enable custom URL | `false` |
|
||||
| `scheduler.admissionWebhook.customURL.host` | Custom URL host | `127.0.0.1` |
|
||||
| `scheduler.admissionWebhook.customURL.port` | Custom URL port | `31998` |
|
||||
| `scheduler.admissionWebhook.customURL.path` | Custom URL path | `/webhook` |
|
||||
| `scheduler.admissionWebhook.reinvocationPolicy` | Reinvocation policy | `Never` |
|
||||
| `scheduler.admissionWebhook.failurePolicy` | Failure policy | `Ignore` |
|
||||
|
||||
### TLS Certificate Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `scheduler.certManager.enabled` | Whether to use cert-manager to generate self-signed certificates | `false` |
|
||||
| `scheduler.patch.enabled` | Whether to use kube-webhook-certgen to generate self-signed certificates | `true` |
|
||||
| `scheduler.patch.image.registry` | Certgen image registry | `"docker.io"` |
|
||||
| `scheduler.patch.image.repository` | Certgen image repository | `"jettech/kube-webhook-certgen"` |
|
||||
| `scheduler.patch.image.tag` | Certgen image tag | `"v1.5.2"` |
|
||||
| `scheduler.patch.image.pullPolicy` | Certgen image pull policy | `IfNotPresent` |
|
||||
| `scheduler.patch.imageNew.registry` | New certgen image registry | `"docker.io"` |
|
||||
| `scheduler.patch.imageNew.repository` | New certgen image repository | `"liangjw/kube-webhook-certgen"` |
|
||||
| `scheduler.patch.imageNew.tag` | New certgen image tag | `"v1.1.1"` |
|
||||
|
||||
### Scheduler Service Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `scheduler.service.type` | Service type | `NodePort` |
|
||||
| `scheduler.service.httpPort` | HTTP port | `443` |
|
||||
| `scheduler.service.schedulerPort` | Scheduler NodePort | `31998` |
|
||||
| `scheduler.service.monitorPort` | Monitor port | `31993` |
|
||||
| `scheduler.service.monitorTargetPort` | Monitor target port | `9395` |
|
||||
|
||||
## Device Plugin Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devicePlugin.image.registry` | Device plugin image registry | `"docker.io"` |
|
||||
| `devicePlugin.image.repository` | Device plugin image repository | `"projecthami/hami"` |
|
||||
| `devicePlugin.image.tag` | Device plugin image tag | `""` |
|
||||
| `devicePlugin.image.pullPolicy` | Device plugin image pull policy | `IfNotPresent` |
|
||||
| `devicePlugin.image.pullSecrets` | Device plugin image pull secrets | `[]` |
|
||||
|
||||
### Monitor Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devicePlugin.monitor.image.registry` | Monitor image registry | `"docker.io"` |
|
||||
| `devicePlugin.monitor.image.repository` | Monitor image repository | `"projecthami/hami"` |
|
||||
| `devicePlugin.monitor.image.tag` | Monitor image tag | `""` |
|
||||
| `devicePlugin.monitor.image.pullPolicy` | Monitor image pull policy | `IfNotPresent` |
|
||||
| `devicePlugin.monitor.image.pullSecrets` | Monitor image pull secrets | `[]` |
|
||||
| `devicePlugin.monitor.ctrPath` | Container path | `/usr/local/vgpu/containers` |
|
||||
| `devicePlugin.monitor.extraArgs` | Monitor extra arguments | `["-v=4"]` |
|
||||
| `devicePlugin.monitor.extraEnvs` | Monitor extra environments | `{}` |
|
||||
|
||||
### Device Plugin Other Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devicePlugin.deviceSplitCount` | Integer type, default value: 10. Maximum number of tasks assigned to a single GPU device | `10` |
|
||||
| `devicePlugin.deviceMemoryScaling` | Device memory scaling ratio | `1` |
|
||||
| `devicePlugin.deviceCoreScaling` | Device core scaling ratio | `1` |
|
||||
| `devicePlugin.runtimeClassName` | Runtime class name | `""` |
|
||||
| `devicePlugin.createRuntimeClass` | Whether to create runtime class | `false` |
|
||||
| `devicePlugin.migStrategy` | String type, "none" means ignore MIG functionality, "mixed" means allocate MIG devices through independent resources | `"none"` |
|
||||
| `devicePlugin.disablecorelimit` | String type, "true" means disable core limit, "false" means enable core limit | `"false"` |
|
||||
| `devicePlugin.passDeviceSpecsEnabled` | Whether to enable passing device specs | `false` |
|
||||
| `devicePlugin.extraArgs` | Device plugin extra arguments | `["-v=4"]` |
|
||||
| `devicePlugin.nodeConfiguration.config` | Node configuration for device plugin by json | An example of default configuration. |
|
||||
| `devicePlugin.nodeConfiguration.externalConfigName` | Node configuration for device plugin by external congimap | `""` |
|
||||
| `devicePlugin.extraEnvs` | Device plugin extra environments | `{}` |
|
||||
|
||||
### Device Plugin Service Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devicePlugin.service.type` | Service type | `NodePort` |
|
||||
| `devicePlugin.service.httpPort` | HTTP port | `31992` |
|
||||
|
||||
### Device Plugin Deployment Configuration
|
||||
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devicePlugin.pluginPath` | Plugin path | `/var/lib/kubelet/device-plugins` |
|
||||
| `devicePlugin.libPath` | Library path | `/usr/local/vgpu` |
|
||||
| `devicePlugin.nvidiaNodeSelector` | NVIDIA node selector | `{"gpu": "on"}` |
|
||||
| `devicePlugin.updateStrategy.type` | Update strategy type | `RollingUpdate` |
|
||||
| `devicePlugin.updateStrategy.rollingUpdate.maxUnavailable` | Maximum unavailable count | `1` |
|
||||
|
||||
## Device Configuration
|
||||
|
||||
### AWS Neuron
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devices.awsneuron.customresources` | Custom resources | `["aws.amazon.com/neuron", "aws.amazon.com/neuroncore"]` |
|
||||
|
||||
### Kunlunxin
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devices.kunlun.enabled` | Whether to enable | `true` |
|
||||
| `devices.kunlun.customresources` | Custom resources | `["kunlunxin.com/xpu"]` |
|
||||
|
||||
### Mthreads
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devices.mthreads.enabled` | Whether to enable | `true` |
|
||||
| `devices.mthreads.customresources` | Custom resources | `["mthreads.com/vgpu"]` |
|
||||
|
||||
### NVIDIA
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devices.nvidia.gpuCorePolicy` | GPU core policy | `default` |
|
||||
| `devices.nvidia.libCudaLogLevel` | CUDA library log level | `1` |
|
||||
|
||||
### Huawei Ascend
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devices.ascend.enabled` | Whether to enable | `false` |
|
||||
| `devices.ascend.image` | Image | `""` |
|
||||
| `devices.ascend.imagePullPolicy` | Image pull policy | `IfNotPresent` |
|
||||
| `devices.ascend.extraArgs` | Extra arguments | `[]` |
|
||||
| `devices.ascend.nodeSelector` | Node selector | `{"ascend": "on"}` |
|
||||
| `devices.ascend.tolerations` | Tolerations | `[]` |
|
||||
| `devices.ascend.customresources` | Custom resources | `["huawei.com/Ascend910A", "huawei.com/Ascend910A-memory", ...]` |
|
||||
|
||||
### Iluvatar
|
||||
| Parameter | Description | Default Value |
|
||||
|-----------|-------------|---------------|
|
||||
| `devices.iluvatar.enabled` | Whether to enable | `false` |
|
||||
| `devices.iluvatar.customresources` | Custom resources | `["iluvatar.ai/BI-V150-vgpu", "iluvatar.ai/BI-V150.vMem","iluvatar.ai/BI-V150.vCore", ...]` |
|
||||
3
packages/system/hami/charts/hami/templates/NOTES.txt
Normal file
3
packages/system/hami/charts/hami/templates/NOTES.txt
Normal file
|
|
@ -0,0 +1,3 @@
|
|||
** Please be patient while the chart is being deployed **
|
||||
Resource name: {{ .Values.resourceName }}
|
||||
|
||||
49
packages/system/hami/charts/hami/templates/_commons.tpl
Normal file
49
packages/system/hami/charts/hami/templates/_commons.tpl
Normal file
|
|
@ -0,0 +1,49 @@
|
|||
{{/*
|
||||
Return the proper image name
|
||||
{{ include "common.images.image" ( dict "imageRoot" .Values.path.to.the.image "global" $) }}
|
||||
*/}}
|
||||
{{- define "common.images.image" -}}
|
||||
{{- $registryName := .imageRoot.registry -}}
|
||||
{{- $repositoryName := .imageRoot.repository -}}
|
||||
{{- $tag := .imageRoot.tag | toString -}}
|
||||
{{- if .global }}
|
||||
{{- if .global.imageRegistry }}
|
||||
{{- $registryName = .global.imageRegistry -}}
|
||||
{{- end -}}
|
||||
{{- end -}}
|
||||
{{- if .tag }}
|
||||
{{- $tag = .tag | toString -}}
|
||||
{{- end -}}
|
||||
{{- if $registryName }}
|
||||
{{- printf "%s/%s:%s" $registryName $repositoryName $tag -}}
|
||||
{{- else -}}
|
||||
{{- printf "%s:%s" $repositoryName $tag -}}
|
||||
{{- end -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
Return the proper Docker Image Registry Secret Names (deprecated: use common.images.renderPullSecrets instead)
|
||||
{{ include "common.images.pullSecrets" ( dict "images" (list .Values.path.to.the.image1, .Values.path.to.the.image2) "global" .Values.global) }}
|
||||
*/}}
|
||||
{{- define "common.images.pullSecrets" -}}
|
||||
{{- $pullSecrets := list }}
|
||||
|
||||
{{- if .global }}
|
||||
{{- range .global.imagePullSecrets -}}
|
||||
{{- $pullSecrets = append $pullSecrets . -}}
|
||||
{{- end -}}
|
||||
{{- end -}}
|
||||
|
||||
{{- range .images -}}
|
||||
{{- range .pullSecrets -}}
|
||||
{{- $pullSecrets = append $pullSecrets . -}}
|
||||
{{- end -}}
|
||||
{{- end -}}
|
||||
|
||||
{{- if (not (empty $pullSecrets)) }}
|
||||
imagePullSecrets:
|
||||
{{- range $pullSecrets }}
|
||||
- name: {{ . }}
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- end -}}
|
||||
163
packages/system/hami/charts/hami/templates/_helpers.tpl
Normal file
163
packages/system/hami/charts/hami/templates/_helpers.tpl
Normal file
|
|
@ -0,0 +1,163 @@
|
|||
{{/*
|
||||
Expand the name of the chart.
|
||||
*/}}
|
||||
{{- define "hami-vgpu.name" -}}
|
||||
{{- default .Chart.Name .Values.nameOverride | trunc 63 | trimSuffix "-" -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
Create a default fully qualified app name.
|
||||
We truncate at 63 chars because some Kubernetes name fields are limited to this (by the DNS naming spec).
|
||||
If release name contains chart name it will be used as a full name.
|
||||
*/}}
|
||||
{{- define "hami-vgpu.fullname" -}}
|
||||
{{- if .Values.fullnameOverride -}}
|
||||
{{- .Values.fullnameOverride | trunc 63 | trimSuffix "-" -}}
|
||||
{{- else }}
|
||||
{{- $name := default .Chart.Name .Values.nameOverride -}}
|
||||
{{- if contains $name .Release.Name }}
|
||||
{{- .Release.Name | trunc 63 | trimSuffix "-" -}}
|
||||
{{- else -}}
|
||||
{{- printf "%s-%s" .Release.Name $name | trunc 63 | trimSuffix "-" -}}
|
||||
{{- end -}}
|
||||
{{- end -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
Allow the release namespace to be overridden for multi-namespace deployments in combined charts
|
||||
*/}}
|
||||
{{- define "hami-vgpu.namespace" -}}
|
||||
{{- if .Values.namespaceOverride -}}
|
||||
{{- .Values.namespaceOverride -}}
|
||||
{{- else -}}
|
||||
{{- .Release.Namespace -}}
|
||||
{{- end -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
The app name for Scheduler
|
||||
*/}}
|
||||
{{- define "hami-vgpu.scheduler" -}}
|
||||
{{- printf "%s-scheduler" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
The app name for DevicePlugin
|
||||
*/}}
|
||||
{{- define "hami-vgpu.device-plugin" -}}
|
||||
{{- printf "%s-device-plugin" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
The app name for MockDevicePlugin
|
||||
*/}}
|
||||
{{- define "hami-vgpu.mock-device-plugin" -}}
|
||||
{{- printf "%s-mock-device-plugin" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
The tls secret name for Scheduler
|
||||
*/}}
|
||||
{{- define "hami-vgpu.scheduler.tls" -}}
|
||||
{{- printf "%s-scheduler-tls" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
The webhook name
|
||||
*/}}
|
||||
{{- define "hami-vgpu.scheduler.webhook" -}}
|
||||
{{- printf "%s-webhook" ( include "hami-vgpu.fullname" . ) | trunc 63 | trimSuffix "-" -}}
|
||||
{{- end -}}
|
||||
|
||||
{{/*
|
||||
Create chart name and version as used by the chart label.
|
||||
*/}}
|
||||
{{- define "hami-vgpu.chart" -}}
|
||||
{{- printf "%s-%s" .Chart.Name .Chart.Version | replace "+" "_" | trunc 63 | trimSuffix "-" }}
|
||||
{{- end }}
|
||||
|
||||
{{/*
|
||||
Common labels
|
||||
*/}}
|
||||
{{- define "hami-vgpu.labels" -}}
|
||||
helm.sh/chart: {{ include "hami-vgpu.chart" . }}
|
||||
{{ include "hami-vgpu.selectorLabels" . }}
|
||||
{{- if .Chart.AppVersion }}
|
||||
app.kubernetes.io/version: {{ .Chart.AppVersion | quote }}
|
||||
{{- end }}
|
||||
app.kubernetes.io/managed-by: {{ .Release.Service }}
|
||||
{{- end }}
|
||||
|
||||
{{/*
|
||||
Selector labels
|
||||
*/}}
|
||||
{{- define "hami-vgpu.selectorLabels" -}}
|
||||
app.kubernetes.io/name: {{ include "hami-vgpu.name" . }}
|
||||
app.kubernetes.io/instance: {{ .Release.Name }}
|
||||
{{- end }}
|
||||
|
||||
|
||||
{{/*
|
||||
Resolve the tag for kubeScheduler.
|
||||
*/}}
|
||||
{{- define "resolvedKubeSchedulerTag" -}}
|
||||
{{- if .Values.scheduler.kubeScheduler.image.tag }}
|
||||
{{- .Values.scheduler.kubeScheduler.image.tag | trim -}}
|
||||
{{- else }}
|
||||
{{- include "strippedKubeVersion" . | trim -}}
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
|
||||
{{/*
|
||||
Return the stripped Kubernetes version string by removing extra parts after semantic version number.
|
||||
v1.31.1+k3s1 -> v1.31.1
|
||||
v1.30.8-eks-2d5f260 -> v1.30.8
|
||||
v1.31.1 -> v1.31.1
|
||||
*/}}
|
||||
{{- define "strippedKubeVersion" -}}
|
||||
{{ regexReplaceAll "^(v[0-9]+\\.[0-9]+\\.[0-9]+)(.*)$" .Capabilities.KubeVersion.Version "$1" }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.scheduler.kubeScheduler.image" -}}
|
||||
{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.kubeScheduler.image "global" .Values.global "tag" (include "resolvedKubeSchedulerTag" .)) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.scheduler.extender.image" -}}
|
||||
{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.extender.image "global" .Values.global "tag" .Values.global.imageTag) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.devicePlugin.image" -}}
|
||||
{{ include "common.images.image" (dict "imageRoot" .Values.devicePlugin.image "global" .Values.global "tag" .Values.global.imageTag) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.mockDevicePlugin.image" -}}
|
||||
{{ include "common.images.image" (dict "imageRoot" .Values.mockDevicePlugin.image "global" .Values.global "tag" .Values.mockDevicePlugin.tag) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.devicePlugin.monitor.image" -}}
|
||||
{{ include "common.images.image" (dict "imageRoot" .Values.devicePlugin.monitor.image "global" .Values.global "tag" .Values.global.imageTag) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.scheduler.patch.image" -}}
|
||||
{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.patch.image "global" .Values.global) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.scheduler.patch.new.image" -}}
|
||||
{{ include "common.images.image" (dict "imageRoot" .Values.scheduler.patch.imageNew "global" .Values.global) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.scheduler.extender.imagePullSecrets" -}}
|
||||
{{ include "common.images.pullSecrets" (dict "images" (list .Values.scheduler.extender.image) "global" .Values.global) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.devicePlugin.imagePullSecrets" -}}
|
||||
{{ include "common.images.pullSecrets" (dict "images" (list .Values.devicePlugin.image) "global" .Values.global) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.scheduler.patch.imagePullSecrets" -}}
|
||||
{{ include "common.images.pullSecrets" (dict "images" (list .Values.scheduler.patch.image) "global" .Values.global) }}
|
||||
{{- end -}}
|
||||
|
||||
{{- define "hami.scheduler.patch.new.imagePullSecrets" -}}
|
||||
{{ include "common.images.pullSecrets" (dict "images" (list .Values.scheduler.patch.imageNew) "global" .Values.global) }}
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,13 @@
|
|||
{{- if and .Values.devicePlugin.enabled (not .Values.devicePlugin.nodeConfiguration.externalConfigName) -}}
|
||||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-device-plugin
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
data:
|
||||
config.json: |
|
||||
{{- .Values.devicePlugin.nodeConfiguration.config | nindent 4 }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,55 @@
|
|||
{{- if .Values.mockDevicePlugin.enabled }}
|
||||
apiVersion: apps/v1
|
||||
kind: DaemonSet
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.mock-device-plugin" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
spec:
|
||||
selector:
|
||||
matchLabels:
|
||||
app.kubernetes.io/component: hami-mock-device-plugin
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 6 }}
|
||||
template:
|
||||
metadata:
|
||||
annotations:
|
||||
scheduler.alpha.kubernetes.io/critical-pod: ""
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-mock-device-plugin
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 8 }}
|
||||
spec:
|
||||
serviceAccountName: {{ include "hami-vgpu.mock-device-plugin" . }}
|
||||
tolerations:
|
||||
- key: CriticalAddonsOnly
|
||||
operator: Exists
|
||||
containers:
|
||||
- image: {{ include "hami.mockDevicePlugin.image" . }}
|
||||
imagePullPolicy: {{ .Values.mockDevicePlugin.image.pullPolicy }}
|
||||
name: hami-mock-dp-cntr
|
||||
env:
|
||||
- name: NODE_NAME
|
||||
valueFrom:
|
||||
fieldRef:
|
||||
fieldPath: spec.nodeName
|
||||
command:
|
||||
- ./k8s-device-plugin
|
||||
- -v=5
|
||||
- --device-config-file=/device-config.yaml
|
||||
volumeMounts:
|
||||
- name: dp
|
||||
mountPath: /var/lib/kubelet/device-plugins
|
||||
- name: sys
|
||||
mountPath: /sys
|
||||
- name: device-config
|
||||
mountPath: /device-config.yaml
|
||||
subPath: device-config.yaml
|
||||
volumes:
|
||||
- name: dp
|
||||
hostPath:
|
||||
path: /var/lib/kubelet/device-plugins
|
||||
- name: sys
|
||||
hostPath:
|
||||
path: /sys
|
||||
- name: device-config
|
||||
configMap:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-device
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,262 @@
|
|||
{{- if .Values.devicePlugin.enabled }}
|
||||
apiVersion: apps/v1
|
||||
kind: DaemonSet
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-device-plugin
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
{{- with .Values.global.labels }}
|
||||
{{- toYaml . | nindent 4 }}
|
||||
{{- end }}
|
||||
{{- if .Values.global.annotations }}
|
||||
annotations: {{ toYaml .Values.global.annotations | nindent 4}}
|
||||
{{- end }}
|
||||
spec:
|
||||
updateStrategy:
|
||||
{{- with .Values.devicePlugin.updateStrategy }}
|
||||
{{- toYaml . | nindent 4 }}
|
||||
{{- end }}
|
||||
selector:
|
||||
matchLabels:
|
||||
app.kubernetes.io/component: hami-device-plugin
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 6 }}
|
||||
template:
|
||||
metadata:
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-device-plugin
|
||||
hami.io/webhook: ignore
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 8 }}
|
||||
annotations:
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
checksum/hami-scheduler-newversion-config: {{ include (print $.Template.BasePath "/scheduler/configmapnew.yaml") . | sha256sum }}
|
||||
{{- else }}
|
||||
checksum/hami-scheduler-config: {{ include (print $.Template.BasePath "/scheduler/configmap.yaml") . | sha256sum }}
|
||||
{{- end }}
|
||||
checksum/hami-scheduler-device-config: {{ include (print $.Template.BasePath "/scheduler/device-configmap.yaml") . | sha256sum }}
|
||||
{{- if .Values.devicePlugin.podAnnotations }}
|
||||
{{- toYaml .Values.devicePlugin.podAnnotations | nindent 8 }}
|
||||
{{- end }}
|
||||
spec:
|
||||
{{- if .Values.devicePlugin.runtimeClassName }}
|
||||
runtimeClassName: {{ .Values.devicePlugin.runtimeClassName }}
|
||||
{{- end }}
|
||||
serviceAccountName: {{ include "hami-vgpu.device-plugin" . }}
|
||||
priorityClassName: system-node-critical
|
||||
hostPID: true
|
||||
hostNetwork: true
|
||||
{{- include "hami.devicePlugin.imagePullSecrets" . | nindent 6 }}
|
||||
{{- if .Values.devicePlugin.gpuOperatorToolkitReady.enabled }}
|
||||
initContainers:
|
||||
- name: toolkit-validation
|
||||
image: {{ include "hami.devicePlugin.image" . }}
|
||||
imagePullPolicy: {{ .Values.devicePlugin.image.pullPolicy }}
|
||||
securityContext:
|
||||
privileged: true
|
||||
runAsUser: 0
|
||||
command: ["sh", "-c"]
|
||||
args:
|
||||
- |
|
||||
echo "Waiting for NVIDIA Toolkit to be ready..."
|
||||
until [ -f {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath }}/toolkit-ready ]; do
|
||||
echo "Waiting for {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath }}/toolkit-ready..."
|
||||
sleep 5
|
||||
done
|
||||
echo "NVIDIA Toolkit is ready!"
|
||||
volumeMounts:
|
||||
- name: nvidia-validations
|
||||
mountPath: {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath | quote }}
|
||||
mountPropagation: HostToContainer
|
||||
readOnly: true
|
||||
{{- end }}
|
||||
containers:
|
||||
- name: device-plugin
|
||||
image: {{ include "hami.devicePlugin.image" . }}
|
||||
imagePullPolicy: {{ .Values.devicePlugin.image.pullPolicy }}
|
||||
lifecycle:
|
||||
postStart:
|
||||
exec:
|
||||
command: ["/bin/sh","-c", {{ printf "/k8s-vgpu/bin/vgpu-init.sh %s/vgpu/" .Values.global.gpuHookPath | quote }}]
|
||||
command:
|
||||
- nvidia-device-plugin
|
||||
- --config-file=/device-config.yaml
|
||||
- --mig-strategy={{ .Values.devicePlugin.migStrategy }}
|
||||
- --disable-core-limit={{ .Values.devicePlugin.disablecorelimit }}
|
||||
{{- range .Values.devicePlugin.extraArgs }}
|
||||
- {{ . }}
|
||||
{{- end }}
|
||||
env:
|
||||
- name: NODE_NAME
|
||||
valueFrom:
|
||||
fieldRef:
|
||||
fieldPath: spec.nodeName
|
||||
- name: NVIDIA_MIG_MONITOR_DEVICES
|
||||
value: all
|
||||
- name: DEVICE_LIST_STRATEGY
|
||||
value: {{ .Values.devicePlugin.deviceListStrategy }}
|
||||
- name: HOOK_PATH
|
||||
value: {{ .Values.global.gpuHookPath }}
|
||||
{{- if typeIs "bool" .Values.devicePlugin.passDeviceSpecsEnabled }}
|
||||
- name: PASS_DEVICE_SPECS
|
||||
value: {{ .Values.devicePlugin.passDeviceSpecsEnabled | quote }}
|
||||
{{- end }}
|
||||
{{- if typeIs "string" .Values.devicePlugin.nvidiaDriverRoot }}
|
||||
- name: NVIDIA_DRIVER_ROOT
|
||||
value: {{ .Values.devicePlugin.nvidiaDriverRoot }}
|
||||
{{- end }}
|
||||
{{- if typeIs "string" .Values.devicePlugin.nvidiaHookPath }}
|
||||
- name: NVIDIA_CDI_HOOK_PATH
|
||||
value: {{ .Values.devicePlugin.nvidiaHookPath }}
|
||||
{{- end }}
|
||||
{{- if typeIs "bool" .Values.devicePlugin.gdrcopyEnabled }}
|
||||
- name: GDRCOPY_ENABLED
|
||||
value: {{ .Values.devicePlugin.gdrcopyEnabled | quote }}
|
||||
{{- end }}
|
||||
{{- if typeIs "bool" .Values.devicePlugin.gdsEnabled }}
|
||||
- name: GDS_ENABLED
|
||||
value: {{ .Values.devicePlugin.gdsEnabled | quote }}
|
||||
{{- end }}
|
||||
{{- if typeIs "bool" .Values.devicePlugin.mofedEnabled }}
|
||||
- name: MOFED_ENABLED
|
||||
value: {{ .Values.devicePlugin.mofedEnabled | quote }}
|
||||
{{- end }}
|
||||
{{- if eq (.Values.scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy | default "spread") "topology-aware" }}
|
||||
- name: ENABLE_TOPOLOGY_SCORE
|
||||
value: "true"
|
||||
{{- end }}
|
||||
{{- with .Values.devicePlugin.extraEnvs }}
|
||||
{{- . | toYaml | nindent 12 }}
|
||||
{{- end }}
|
||||
securityContext:
|
||||
privileged: true
|
||||
allowPrivilegeEscalation: true
|
||||
capabilities:
|
||||
drop: ["ALL"]
|
||||
add: ["SYS_ADMIN"]
|
||||
resources:
|
||||
{{- toYaml .Values.devicePlugin.resources | nindent 12 }}
|
||||
volumeMounts:
|
||||
- name: device-plugin
|
||||
mountPath: /var/lib/kubelet/device-plugins
|
||||
- name: lib
|
||||
mountPath: {{ printf "%s%s" .Values.global.gpuHookPath "/vgpu" }}
|
||||
- name: usrbin
|
||||
mountPath: /usrbin
|
||||
- name: deviceconfig
|
||||
mountPath: /config
|
||||
- name: hosttmp
|
||||
mountPath: /tmp
|
||||
- name: device-config
|
||||
mountPath: /device-config.yaml
|
||||
subPath: device-config.yaml
|
||||
- name: cdi-root
|
||||
mountPath: /var/run/cdi
|
||||
{{- if typeIs "string" .Values.devicePlugin.nvidiaDriverRoot }}
|
||||
# We always mount the driver root at /driver-root in the container.
|
||||
# This is required for CDI detection to work correctly.
|
||||
- name: driver-root
|
||||
mountPath: /driver-root
|
||||
readOnly: true
|
||||
{{- end }}
|
||||
- name: vgpu-monitor
|
||||
image: {{ include "hami.devicePlugin.monitor.image" . }}
|
||||
imagePullPolicy: {{ .Values.devicePlugin.monitor.image.pullPolicy }}
|
||||
command:
|
||||
- "vGPUmonitor"
|
||||
{{- range .Values.devicePlugin.monitor.extraArgs }}
|
||||
- {{ . }}
|
||||
{{- end }}
|
||||
securityContext:
|
||||
allowPrivilegeEscalation: false
|
||||
capabilities:
|
||||
drop: ["ALL"]
|
||||
add: ["SYS_ADMIN"]
|
||||
env:
|
||||
- name: NODE_NAME
|
||||
valueFrom:
|
||||
fieldRef:
|
||||
fieldPath: spec.nodeName
|
||||
- name: NVIDIA_VISIBLE_DEVICES
|
||||
value: "all"
|
||||
- name: NVIDIA_MIG_MONITOR_DEVICES
|
||||
value: "all"
|
||||
- name: HOOK_PATH
|
||||
value: "{{ .Values.global.gpuHookPath }}/vgpu"
|
||||
- name: HAMI_RESYNC_INTERVAL
|
||||
value: {{ .Values.devicePlugin.monitor.resyncInterval | default "5m" | quote }}
|
||||
{{- with .Values.devicePlugin.monitor.extraEnvs }}
|
||||
{{- . | toYaml | nindent 12 }}
|
||||
{{- end }}
|
||||
resources:
|
||||
{{- toYaml .Values.devicePlugin.monitor.resources | nindent 12 }}
|
||||
volumeMounts:
|
||||
- name: ctrs
|
||||
mountPath: {{ .Values.devicePlugin.monitor.ctrPath }}
|
||||
- name: dockers
|
||||
mountPath: /run/docker
|
||||
- name: containerds
|
||||
mountPath: /run/containerd
|
||||
- name: sysinfo
|
||||
mountPath: /sysinfo
|
||||
- name: hostvar
|
||||
mountPath: /hostvar
|
||||
- name: hosttmp
|
||||
mountPath: /tmp
|
||||
volumes:
|
||||
- name: ctrs
|
||||
hostPath:
|
||||
path: {{ .Values.devicePlugin.monitor.ctrPath }}
|
||||
- name: hosttmp
|
||||
hostPath:
|
||||
path: /tmp
|
||||
- name: dockers
|
||||
hostPath:
|
||||
path: /run/docker
|
||||
- name: containerds
|
||||
hostPath:
|
||||
path: /run/containerd
|
||||
- name: device-plugin
|
||||
hostPath:
|
||||
path: {{ .Values.devicePlugin.pluginPath }}
|
||||
- name: lib
|
||||
hostPath:
|
||||
path: {{ .Values.devicePlugin.libPath }}
|
||||
{{- if .Values.devicePlugin.gpuOperatorToolkitReady.enabled }}
|
||||
- name: nvidia-validations
|
||||
hostPath:
|
||||
path: {{ .Values.devicePlugin.gpuOperatorToolkitReady.hostPath }}
|
||||
type: DirectoryOrCreate
|
||||
{{- end }}
|
||||
{{- if typeIs "string" .Values.devicePlugin.nvidiaDriverRoot }}
|
||||
- name: driver-root
|
||||
hostPath:
|
||||
path: {{ .Values.devicePlugin.nvidiaDriverRoot }}
|
||||
type: Directory
|
||||
{{- end }}
|
||||
- name: cdi-root
|
||||
hostPath:
|
||||
path: /var/run/cdi
|
||||
type: DirectoryOrCreate
|
||||
- name: usrbin
|
||||
hostPath:
|
||||
path: /usr/bin
|
||||
- name: sysinfo
|
||||
hostPath:
|
||||
path: /sys
|
||||
- name: hostvar
|
||||
hostPath:
|
||||
path: /var
|
||||
- name: deviceconfig
|
||||
configMap:
|
||||
name: {{ .Values.devicePlugin.nodeConfiguration.externalConfigName | default (include "hami-vgpu.device-plugin" .) }}
|
||||
- name: device-config
|
||||
configMap:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-device
|
||||
{{- if .Values.devicePlugin.nvidiaNodeSelector }}
|
||||
nodeSelector: {{ toYaml .Values.devicePlugin.nvidiaNodeSelector | nindent 8 }}
|
||||
{{- end }}
|
||||
{{- if .Values.devicePlugin.tolerations }}
|
||||
tolerations: {{ toYaml .Values.devicePlugin.tolerations | nindent 8 }}
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,28 @@
|
|||
{{- if .Values.devicePlugin.enabled -}}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRole
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}-monitor
|
||||
rules:
|
||||
- apiGroups:
|
||||
- ""
|
||||
resources:
|
||||
- pods
|
||||
verbs:
|
||||
- get
|
||||
- create
|
||||
- watch
|
||||
- list
|
||||
- update
|
||||
- patch
|
||||
- apiGroups:
|
||||
- ""
|
||||
resources:
|
||||
- nodes
|
||||
verbs:
|
||||
- get
|
||||
- update
|
||||
- list
|
||||
- patch
|
||||
{{- end -}}
|
||||
|
||||
|
|
@ -0,0 +1,17 @@
|
|||
{{- if .Values.devicePlugin.enabled -}}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-device-plugin"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: ClusterRole
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}-monitor
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,29 @@
|
|||
{{- if .Values.devicePlugin.enabled -}}
|
||||
apiVersion: v1
|
||||
kind: Service
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}-monitor
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-device-plugin
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
{{- with .Values.devicePlugin.service.labels }}
|
||||
{{- toYaml . | nindent 4 }}
|
||||
{{- end }}
|
||||
{{- if .Values.devicePlugin.service.annotations }} # Use devicePlugin instead of scheduler
|
||||
annotations: {{ toYaml .Values.devicePlugin.service.annotations | nindent 4 }}
|
||||
{{- end }}
|
||||
spec:
|
||||
type: {{ .Values.devicePlugin.service.type | default "NodePort" }} # Default type is NodePort
|
||||
ports:
|
||||
- name: monitorport
|
||||
port: {{ .Values.devicePlugin.service.httpPort | default 31992 }} # Default HTTP port is 31992
|
||||
targetPort: 9394
|
||||
{{- if eq (.Values.devicePlugin.service.type | default "NodePort") "NodePort" }} # If type is NodePort, set nodePort
|
||||
nodePort: {{ .Values.devicePlugin.service.httpPort | default 31992 }}
|
||||
{{- end }}
|
||||
protocol: TCP
|
||||
selector:
|
||||
app.kubernetes.io/component: hami-device-plugin
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 4 }}
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
{{- if .Values.devicePlugin.enabled -}}
|
||||
apiVersion: v1
|
||||
kind: ServiceAccount
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.device-plugin" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-device-plugin"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,9 @@
|
|||
{{- if and .Values.devicePlugin.enabled .Values.devicePlugin.createRuntimeClass .Values.devicePlugin.runtimeClassName -}}
|
||||
apiVersion: node.k8s.io/v1
|
||||
kind: RuntimeClass
|
||||
metadata:
|
||||
name: {{ .Values.devicePlugin.runtimeClassName }}
|
||||
annotations:
|
||||
helm.sh/hook: pre-install,pre-upgrade
|
||||
handler: nvidia
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,31 @@
|
|||
{{- if .Values.scheduler.admissionWebhook.enabled -}}
|
||||
{{- if .Values.scheduler.certManager.enabled }}
|
||||
apiVersion: cert-manager.io/v1
|
||||
kind: Certificate
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-serving-cert
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
spec:
|
||||
dnsNames:
|
||||
- {{ include "hami-vgpu.scheduler" . }}.{{ include "hami-vgpu.namespace" . }}.svc
|
||||
- {{ include "hami-vgpu.scheduler" . }}.{{ include "hami-vgpu.namespace" . }}.svc.cluster.local
|
||||
issuerRef:
|
||||
kind: Issuer
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-selfsigned-issuer
|
||||
secretName: {{ include "hami-vgpu.scheduler.tls" . }}
|
||||
---
|
||||
apiVersion: cert-manager.io/v1
|
||||
kind: Issuer
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-selfsigned-issuer
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
spec:
|
||||
selfSigned: {}
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,34 @@
|
|||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRole
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-scheduler"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
rules:
|
||||
- apiGroups: [""]
|
||||
resources: ["pods", "configmaps"]
|
||||
verbs: ["get", "list", "watch", "patch"]
|
||||
- apiGroups: [""]
|
||||
resources: ["pods/binding"]
|
||||
verbs: ["create"]
|
||||
- apiGroups: [""]
|
||||
resources: ["nodes"]
|
||||
verbs: ["get", "list", "patch", "watch"]
|
||||
- apiGroups: [""]
|
||||
resources: ["events"]
|
||||
verbs: ["create", "get", "list"]
|
||||
- apiGroups: [""]
|
||||
resources: ["resourcequotas"]
|
||||
verbs: ["get", "list", "watch"]
|
||||
---
|
||||
{{- if .Values.mockDevicePlugin.enabled }}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRole
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.mock-device-plugin" . }}
|
||||
rules:
|
||||
- apiGroups: [""]
|
||||
resources: ["nodes"]
|
||||
verbs: ["get", "update", "list", "patch"]
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,47 @@
|
|||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-kube
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-scheduler"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: ClusterRole
|
||||
name: system:kube-scheduler
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
---
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-volume
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-scheduler"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: ClusterRole
|
||||
name: system:volume-scheduler
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
---
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-scheduler"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: ClusterRole
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
|
|
@ -0,0 +1,142 @@
|
|||
{{- if .Values.scheduler.kubeScheduler.enabled -}}
|
||||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
data:
|
||||
config.json: |
|
||||
{
|
||||
"kind": "Policy",
|
||||
"apiVersion": "v1",
|
||||
"extenders": [
|
||||
{
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
"urlPrefix": "https://127.0.0.1:443",
|
||||
"enableHttps": true,
|
||||
"tlsConfig": {
|
||||
"insecure": true
|
||||
},
|
||||
{{- else }}
|
||||
"urlPrefix": "http://127.0.0.1:80",
|
||||
"enableHttps": false,
|
||||
{{- end }}
|
||||
"filterVerb": "filter",
|
||||
"bindVerb": "bind",
|
||||
"weight": 1,
|
||||
"nodeCacheCapable": true,
|
||||
"httpTimeout": 30000000000,
|
||||
"managedResources": [
|
||||
{{- range .Values.devices.amd.customresources }}
|
||||
{
|
||||
"name": "{{ . }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{{- end }}
|
||||
{{- if .Values.devices.ascend.enabled }}
|
||||
{{- range .Values.devices.ascend.customresources }}
|
||||
{
|
||||
"name": "{{ . }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- if .Values.devices.mthreads.enabled }}
|
||||
{{- range .Values.devices.mthreads.customresources }}
|
||||
{
|
||||
"name": "{{ . }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- if .Values.devices.enflame.enabled }}
|
||||
{{- range .Values.devices.enflame.customresources }}
|
||||
{
|
||||
"name": "{{ . }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- if .Values.devices.kunlun.enabled }}
|
||||
{{- range .Values.devices.kunlun.customresources }}
|
||||
{
|
||||
"name": "{{ . }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- range .Values.devices.awsneuron.customresources }}
|
||||
{
|
||||
"name": "{{ . }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{{- end }}
|
||||
{{- if .Values.devices.iluvatar.enabled }}
|
||||
{{- range .Values.devices.iluvatar.customresources }}
|
||||
{
|
||||
"name": "{{ . }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{
|
||||
"name": "{{ .Values.resourceName }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.resourceMem }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.resourceCores }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.resourceMemPercentage }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.resourcePriority }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.mluResourceName }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.dcuResourceName }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.dcuResourceMem }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.dcuResourceCores }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "metax-tech.com/gpu",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.metaxResourceName }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.metaxResourceCore }}",
|
||||
"ignoredByScheduler": true
|
||||
},
|
||||
{
|
||||
"name": "{{ .Values.metaxResourceMem }}",
|
||||
"ignoredByScheduler": true
|
||||
}
|
||||
],
|
||||
"ignoreable": false
|
||||
}
|
||||
]
|
||||
}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,102 @@
|
|||
{{- if .Values.scheduler.kubeScheduler.enabled -}}
|
||||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-newversion
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
data:
|
||||
config.yaml: |
|
||||
{{- if gt (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 25}}
|
||||
apiVersion: kubescheduler.config.k8s.io/v1
|
||||
{{- else }}
|
||||
apiVersion: kubescheduler.config.k8s.io/v1beta2
|
||||
{{- end }}
|
||||
kind: KubeSchedulerConfiguration
|
||||
leaderElection:
|
||||
leaderElect: false
|
||||
profiles:
|
||||
- schedulerName: {{ .Values.schedulerName }}
|
||||
extenders:
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
- urlPrefix: "https://127.0.0.1:443"
|
||||
enableHTTPS: true
|
||||
tlsConfig:
|
||||
insecure: true
|
||||
{{- else }}
|
||||
- urlPrefix: "http://127.0.0.1:80"
|
||||
enableHTTPS: false
|
||||
{{- end }}
|
||||
filterVerb: filter
|
||||
bindVerb: bind
|
||||
nodeCacheCapable: true
|
||||
weight: 1
|
||||
httpTimeout: 30s
|
||||
managedResources:
|
||||
- name: {{ .Values.resourceName }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.resourceMem }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.resourceCores }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.resourceMemPercentage }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.resourcePriority }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.mluResourceName }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.dcuResourceName }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.dcuResourceMem }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.dcuResourceCores }}
|
||||
ignoredByScheduler: true
|
||||
- name: "metax-tech.com/gpu"
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.metaxResourceName }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.metaxResourceCore }}
|
||||
ignoredByScheduler: true
|
||||
- name: {{ .Values.metaxResourceMem }}
|
||||
ignoredByScheduler: true
|
||||
{{- if .Values.devices.ascend.enabled }}
|
||||
{{- range .Values.devices.ascend.customresources }}
|
||||
- name: {{ . }}
|
||||
ignoredByScheduler: true
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- if .Values.devices.mthreads.enabled }}
|
||||
{{- range .Values.devices.mthreads.customresources }}
|
||||
- name: {{ . }}
|
||||
ignoredByScheduler: true
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- if .Values.devices.enflame.enabled }}
|
||||
{{- range .Values.devices.enflame.customresources }}
|
||||
- name: {{ . }}
|
||||
ignoredByScheduler: true
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- if .Values.devices.kunlun.enabled }}
|
||||
{{- range .Values.devices.kunlun.customresources }}
|
||||
- name: {{ . }}
|
||||
ignoredByScheduler: true
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- range .Values.devices.awsneuron.customresources }}
|
||||
- name: {{ . }}
|
||||
ignoredByScheduler: true
|
||||
{{- end }}
|
||||
{{- if .Values.devices.iluvatar.enabled }}
|
||||
{{- range .Values.devices.iluvatar.customresources }}
|
||||
- name: {{ . }}
|
||||
ignoredByScheduler: true
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
{{- range .Values.devices.amd.customresources }}
|
||||
- name: {{ . }}
|
||||
ignoredByScheduler: true
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,225 @@
|
|||
apiVersion: apps/v1
|
||||
kind: Deployment
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
{{- with .Values.global.labels }}
|
||||
{{- toYaml . | nindent 4 }}
|
||||
{{- end }}
|
||||
{{- if .Values.global.annotations }}
|
||||
annotations: {{ toYaml .Values.global.annotations | nindent 4}}
|
||||
{{- end }}
|
||||
spec:
|
||||
{{- if .Values.scheduler.leaderElect }}
|
||||
replicas: {{ .Values.scheduler.replicas }}
|
||||
{{- else }}
|
||||
replicas: 1
|
||||
{{- end }}
|
||||
selector:
|
||||
matchLabels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 6 }}
|
||||
template:
|
||||
metadata:
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 8 }}
|
||||
hami.io/webhook: ignore
|
||||
annotations:
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
checksum/hami-scheduler-newversion-config: {{ include (print $.Template.BasePath "/scheduler/configmapnew.yaml") . | sha256sum }}
|
||||
{{- else }}
|
||||
checksum/hami-scheduler-config: {{ include (print $.Template.BasePath "/scheduler/configmap.yaml") . | sha256sum }}
|
||||
{{- end }}
|
||||
checksum/hami-scheduler-device-config: {{ include (print $.Template.BasePath "/scheduler/device-configmap.yaml") . | sha256sum }}
|
||||
{{- if .Values.scheduler.podAnnotations }}
|
||||
{{- toYaml .Values.scheduler.podAnnotations | nindent 8 }}
|
||||
{{- end }}
|
||||
spec:
|
||||
serviceAccountName: {{ include "hami-vgpu.scheduler" . }}
|
||||
priorityClassName: system-node-critical
|
||||
{{- include "hami.scheduler.extender.imagePullSecrets" . | nindent 6 }}
|
||||
containers:
|
||||
{{- if .Values.scheduler.kubeScheduler.enabled }}
|
||||
- name: kube-scheduler
|
||||
image: {{ include "hami.scheduler.kubeScheduler.image" . }}
|
||||
imagePullPolicy: {{ .Values.scheduler.kubeScheduler.image.pullPolicy }}
|
||||
command:
|
||||
- kube-scheduler
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
{{- range .Values.scheduler.kubeScheduler.extraNewArgs }}
|
||||
- {{ . }}
|
||||
{{- end }}
|
||||
{{- else }}
|
||||
- --scheduler-name={{ .Values.schedulerName }}
|
||||
{{- range .Values.scheduler.kubeScheduler.extraArgs }}
|
||||
- {{ . }}
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
- --leader-elect={{ .Values.scheduler.leaderElect }}
|
||||
- --leader-elect-resource-name={{ .Values.schedulerName }}
|
||||
- --leader-elect-resource-namespace={{ include "hami-vgpu.namespace" . }}
|
||||
resources:
|
||||
{{- toYaml .Values.scheduler.kubeScheduler.resources | nindent 12 }}
|
||||
volumeMounts:
|
||||
- name: scheduler-config
|
||||
mountPath: /config
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.livenessProbe }}
|
||||
livenessProbe:
|
||||
failureThreshold: 8
|
||||
httpGet:
|
||||
path: /healthz
|
||||
port: 10259
|
||||
scheme: HTTPS
|
||||
initialDelaySeconds: 10
|
||||
periodSeconds: 10
|
||||
successThreshold: 1
|
||||
timeoutSeconds: 15
|
||||
{{- end }}
|
||||
- name: vgpu-scheduler-extender
|
||||
image: {{ include "hami.scheduler.extender.image" . }}
|
||||
imagePullPolicy: {{ .Values.scheduler.extender.image.pullPolicy }}
|
||||
env:
|
||||
{{- if .Values.scheduler.nodeLockExpire }}
|
||||
- name: HAMI_NODELOCK_EXPIRE
|
||||
value: "{{ .Values.scheduler.nodeLockExpire }}"
|
||||
{{- end }}
|
||||
{{- if .Values.global.managedNodeSelectorEnable }}
|
||||
{{- range $key, $value := .Values.global.managedNodeSelector }}
|
||||
- name: NODE_SELECTOR_{{ $key | upper | replace "-" "_" }}
|
||||
value: "{{ $value }}"
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
command:
|
||||
- scheduler
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
- --http_bind=0.0.0.0:443
|
||||
- --cert_file=/tls/tls.crt
|
||||
- --key_file=/tls/tls.key
|
||||
{{- else }}
|
||||
- --http_bind=0.0.0.0:80
|
||||
{{- end }}
|
||||
- --scheduler-name={{ .Values.schedulerName }}
|
||||
- --metrics-bind-address={{ .Values.scheduler.metricsBindAddress }}
|
||||
- --node-scheduler-policy={{ .Values.scheduler.defaultSchedulerPolicy.nodeSchedulerPolicy }}
|
||||
- --gpu-scheduler-policy={{ .Values.scheduler.defaultSchedulerPolicy.gpuSchedulerPolicy }}
|
||||
- --force-overwrite-default-scheduler={{ .Values.scheduler.forceOverwriteDefaultScheduler}}
|
||||
- --device-config-file=/device-config.yaml
|
||||
- --leader-elect={{ .Values.scheduler.leaderElect }}
|
||||
- --leader-elect-resource-name={{ .Values.schedulerName }}
|
||||
- --leader-elect-resource-namespace={{ include "hami-vgpu.namespace" . }}
|
||||
{{- if .Values.devices.ascend.enabled }}
|
||||
- --enable-ascend=true
|
||||
{{- end }}
|
||||
{{- if .Values.devices.iluvatar.enabled }}
|
||||
- --enable-iluvatar=true
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.nodeLabelSelector }}
|
||||
- --node-label-selector={{- $first := true -}}
|
||||
{{- range $key, $value := .Values.scheduler.nodeLabelSelector -}}
|
||||
{{- if not $first }},{{ end -}}
|
||||
{{- $key }}={{ $value -}}
|
||||
{{- $first = false -}}
|
||||
{{- end -}}
|
||||
{{- end }}
|
||||
{{- range .Values.scheduler.extender.extraArgs }}
|
||||
- {{ . }}
|
||||
{{- end }}
|
||||
ports:
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
- name: https
|
||||
containerPort: 443
|
||||
protocol: TCP
|
||||
{{- else }}
|
||||
- name: http
|
||||
containerPort: 80
|
||||
protocol: TCP
|
||||
{{- end }}
|
||||
- name: metrics
|
||||
containerPort: {{ last (splitList ":" .Values.scheduler.metricsBindAddress) | int }}
|
||||
protocol: TCP
|
||||
resources:
|
||||
{{- toYaml .Values.scheduler.extender.resources | nindent 12 }}
|
||||
volumeMounts:
|
||||
- name: device-config
|
||||
mountPath: /device-config.yaml
|
||||
subPath: device-config.yaml
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
- name: tls-config
|
||||
mountPath: /tls
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.livenessProbe }}
|
||||
livenessProbe:
|
||||
httpGet:
|
||||
path: /healthz
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
port: https
|
||||
scheme: HTTPS
|
||||
{{- else }}
|
||||
port: http
|
||||
scheme: HTTP
|
||||
{{- end }}
|
||||
initialDelaySeconds: 10
|
||||
periodSeconds: 10
|
||||
failureThreshold: 3
|
||||
timeoutSeconds: 5
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.leaderElect }}
|
||||
readinessProbe:
|
||||
httpGet:
|
||||
path: /readyz
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
port: https
|
||||
scheme: HTTPS
|
||||
{{- else }}
|
||||
port: http
|
||||
scheme: HTTP
|
||||
{{- end }}
|
||||
initialDelaySeconds: 10
|
||||
periodSeconds: 10
|
||||
failureThreshold: 3
|
||||
timeoutSeconds: 5
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.leaderElect }}
|
||||
affinity:
|
||||
podAntiAffinity:
|
||||
requiredDuringSchedulingIgnoredDuringExecution:
|
||||
- labelSelector:
|
||||
matchExpressions:
|
||||
- key: app.kubernetes.io/component
|
||||
operator: In
|
||||
values:
|
||||
- hami-scheduler
|
||||
topologyKey: "kubernetes.io/hostname"
|
||||
{{- end }}
|
||||
volumes:
|
||||
{{- if .Values.scheduler.admissionWebhook.enabled }}
|
||||
- name: tls-config
|
||||
secret:
|
||||
secretName: {{ template "hami-vgpu.scheduler.tls" . }}
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.kubeScheduler.enabled }}
|
||||
- name: scheduler-config
|
||||
configMap:
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
name: {{ template "hami-vgpu.scheduler" . }}-newversion
|
||||
{{- else }}
|
||||
name: {{ template "hami-vgpu.scheduler" . }}
|
||||
{{- end }}
|
||||
{{- end }}
|
||||
- name: device-config
|
||||
configMap:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-device
|
||||
{{- if .Values.scheduler.nodeSelector }}
|
||||
nodeSelector: {{ toYaml .Values.scheduler.nodeSelector | nindent 8 }}
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.tolerations }}
|
||||
tolerations: {{ toYaml .Values.scheduler.tolerations | nindent 8 }}
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.nodeName }}
|
||||
nodeName: {{ .Values.scheduler.nodeName }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,408 @@
|
|||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}-device
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
data:
|
||||
device-config.yaml: |-
|
||||
{{- if .Files.Glob "files/device-config.yaml" }}
|
||||
{{- .Files.Get "files/device-config.yaml" | nindent 4}}
|
||||
{{- else }}
|
||||
nvidia:
|
||||
resourceCountName: {{ .Values.resourceName }}
|
||||
resourceMemoryName: {{ .Values.resourceMem }}
|
||||
resourceMemoryPercentageName: {{ .Values.resourceMemPercentage }}
|
||||
resourceCoreName: {{ .Values.resourceCores }}
|
||||
resourcePriorityName: {{ .Values.resourcePriority }}
|
||||
overwriteEnv: false
|
||||
defaultMemory: 0
|
||||
defaultCores: 0
|
||||
defaultGPUNum: 1
|
||||
memoryFactor: 1
|
||||
deviceSplitCount: {{ .Values.devicePlugin.deviceSplitCount }}
|
||||
deviceMemoryScaling: {{ .Values.devicePlugin.deviceMemoryScaling }}
|
||||
deviceCoreScaling: {{ .Values.devicePlugin.deviceCoreScaling }}
|
||||
gpuCorePolicy: {{ .Values.devices.nvidia.gpuCorePolicy }}
|
||||
libCudaLogLevel: {{ .Values.devices.nvidia.libCudaLogLevel }}
|
||||
runtimeClassName: "{{ .Values.devicePlugin.runtimeClassName }}"
|
||||
knownMigGeometries:
|
||||
- models: [ "A30" ]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.6gb
|
||||
core: 25
|
||||
memory: 6144
|
||||
count: 4
|
||||
-
|
||||
- name: 2g.12gb
|
||||
core: 50
|
||||
memory: 12288
|
||||
count: 2
|
||||
-
|
||||
- name: 4g.24gb
|
||||
core: 100
|
||||
memory: 24576
|
||||
count: 1
|
||||
- models: [ "A100-SXM4-40GB", "A100-40GB-PCIe", "A100-PCIE-40GB"]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.5gb
|
||||
core: 14
|
||||
memory: 5120
|
||||
count: 7
|
||||
-
|
||||
- name: 1g.5gb
|
||||
core: 14
|
||||
memory: 5120
|
||||
count: 1
|
||||
- name: 2g.10gb
|
||||
core: 28
|
||||
memory: 10240
|
||||
count: 3
|
||||
-
|
||||
- name: 3g.20gb
|
||||
core: 42
|
||||
memory: 20480
|
||||
count: 2
|
||||
-
|
||||
- name: 7g.40gb
|
||||
core: 100
|
||||
memory: 40960
|
||||
count: 1
|
||||
- models: [ "A100-SXM4-80GB", "A100-80GB-PCIe", "A100-PCIE-80GB"]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.10gb
|
||||
core: 14
|
||||
memory: 10240
|
||||
count: 7
|
||||
-
|
||||
- name: 1g.10gb
|
||||
core: 14
|
||||
memory: 10240
|
||||
count: 1
|
||||
- name: 2g.20gb
|
||||
core: 28
|
||||
memory: 20480
|
||||
count: 3
|
||||
-
|
||||
- name: 3g.40gb
|
||||
core: 42
|
||||
memory: 40960
|
||||
count: 2
|
||||
-
|
||||
- name: 7g.79gb
|
||||
core: 100
|
||||
memory: 80896
|
||||
count: 1
|
||||
- models: [ "H100-PCIE-80GB", "H100-SXM5-80GB"]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.10gb
|
||||
core: 14
|
||||
memory: 10240
|
||||
count: 7
|
||||
-
|
||||
- name: 1g.10gb
|
||||
core: 14
|
||||
memory: 10240
|
||||
count: 1
|
||||
- name: 2g.20gb
|
||||
core: 28
|
||||
memory: 20480
|
||||
count: 3
|
||||
-
|
||||
- name: 3g.40gb
|
||||
core: 42
|
||||
memory: 40960
|
||||
count: 2
|
||||
-
|
||||
- name: 7g.80gb
|
||||
core: 100
|
||||
memory: 81920
|
||||
count: 1
|
||||
- models: [ "H100-PCIE-94GB", "H100-SXM5-94GB"]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.12gb
|
||||
core: 14
|
||||
memory: 12288
|
||||
count: 7
|
||||
-
|
||||
- name: 1g.12gb
|
||||
core: 14
|
||||
memory: 12288
|
||||
count: 1
|
||||
- name: 2g.24gb
|
||||
core: 28
|
||||
memory: 24576
|
||||
count: 3
|
||||
-
|
||||
- name: 3g.47gb
|
||||
core: 42
|
||||
memory: 48128
|
||||
count: 2
|
||||
-
|
||||
- name: 7g.94gb
|
||||
core: 100
|
||||
memory: 96256
|
||||
count: 1
|
||||
- models: [ "H20", "H100 on GH200"]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.12gb
|
||||
core: 14
|
||||
memory: 12288
|
||||
count: 7
|
||||
-
|
||||
- name: 1g.12gb
|
||||
core: 14
|
||||
memory: 12288
|
||||
count: 1
|
||||
- name: 2g.24gb
|
||||
core: 28
|
||||
memory: 24576
|
||||
count: 3
|
||||
-
|
||||
- name: 3g.48gb
|
||||
core: 42
|
||||
memory: 49152
|
||||
count: 2
|
||||
-
|
||||
- name: 7g.96gb
|
||||
core: 100
|
||||
memory: 98304
|
||||
count: 1
|
||||
- models: [ "H200 NVL", "H200-SXM5"]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.18gb
|
||||
core: 14
|
||||
memory: 18432
|
||||
count: 7
|
||||
-
|
||||
- name: 1g.18gb
|
||||
core: 14
|
||||
memory: 18432
|
||||
count: 1
|
||||
- name: 2g.35gb
|
||||
core: 28
|
||||
memory: 35840
|
||||
count: 3
|
||||
-
|
||||
- name: 3g.71gb
|
||||
core: 42
|
||||
memory: 72704
|
||||
count: 2
|
||||
-
|
||||
- name: 7g.141gb
|
||||
core: 100
|
||||
memory: 144384
|
||||
count: 1
|
||||
- models: [ "B200" ]
|
||||
allowedGeometries:
|
||||
-
|
||||
- name: 1g.23gb
|
||||
core: 14
|
||||
memory: 23552
|
||||
count: 7
|
||||
-
|
||||
- name: 1g.23gb
|
||||
core: 14
|
||||
memory: 23552
|
||||
count: 1
|
||||
- name: 2g.45gb
|
||||
core: 28
|
||||
memory: 46080
|
||||
count: 3
|
||||
-
|
||||
- name: 3g.90gb
|
||||
core: 42
|
||||
memory: 92160
|
||||
count: 2
|
||||
-
|
||||
- name: 7g.180gb
|
||||
core: 100
|
||||
memory: 184320
|
||||
count: 1
|
||||
cambricon:
|
||||
resourceCountName: {{ .Values.mluResourceName }}
|
||||
resourceMemoryName: {{ .Values.mluResourceMem }}
|
||||
resourceCoreName: {{ .Values.mluResourceCores }}
|
||||
hygon:
|
||||
resourceCountName: {{ .Values.dcuResourceName }}
|
||||
resourceMemoryName: {{ .Values.dcuResourceMem }}
|
||||
resourceCoreName: {{ .Values.dcuResourceCores }}
|
||||
memoryFactor: 1
|
||||
metax:
|
||||
resourceCountName: "metax-tech.com/gpu"
|
||||
resourceVCountName: {{ .Values.metaxResourceName }}
|
||||
resourceVMemoryName: {{ .Values.metaxResourceMem }}
|
||||
resourceVCoreName: {{ .Values.metaxResourceCore }}
|
||||
sgpuTopologyAware: {{ .Values.metaxsGPUTopologyAware }}
|
||||
enflame:
|
||||
resourceNameGCU: "enflame.com/gcu"
|
||||
resourceNameVGCU: {{ .Values.enflameResourceNameVGCU }}
|
||||
resourceNameVGCUPercentage: {{ .Values.enflameResourceNameVGCUPercentage }}
|
||||
mthreads:
|
||||
resourceCountName: "mthreads.com/vgpu"
|
||||
resourceMemoryName: "mthreads.com/sgpu-memory"
|
||||
resourceCoreName: "mthreads.com/sgpu-core"
|
||||
iluvatars:
|
||||
- chipName: MR-V100
|
||||
commonWord: MR-V100
|
||||
resourceCountName: iluvatar.ai/MR-V100-vgpu
|
||||
resourceMemoryName: iluvatar.ai/MR-V100.vMem
|
||||
resourceCoreName: iluvatar.ai/MR-V100.vCore
|
||||
- chipName: MR-V50
|
||||
commonWord: MR-V50
|
||||
resourceCountName: iluvatar.ai/MR-V50-vgpu
|
||||
resourceMemoryName: iluvatar.ai/MR-V50.vMem
|
||||
resourceCoreName: iluvatar.ai/MR-V50.vCore
|
||||
- chipName: BI-V150
|
||||
commonWord: BI-V150
|
||||
resourceCountName: iluvatar.ai/BI-V150-vgpu
|
||||
resourceMemoryName: iluvatar.ai/BI-V150.vMem
|
||||
resourceCoreName: iluvatar.ai/BI-V150.vCore
|
||||
- chipName: BI-V100
|
||||
commonWord: BI-V100
|
||||
resourceCountName: iluvatar.ai/BI-V100-vgpu
|
||||
resourceMemoryName: iluvatar.ai/BI-V100.vMem
|
||||
resourceCoreName: iluvatar.ai/BI-V100.vCore
|
||||
kunlun:
|
||||
resourceCountName: {{ .Values.kunlunResourceName }}
|
||||
resourceVCountName: {{ .Values.kunlunResourceVCountName }}
|
||||
resourceVMemoryName: {{ .Values.kunlunResourceVMemoryName }}
|
||||
awsneuron:
|
||||
resourceCountName: "aws.amazon.com/neuron"
|
||||
resourceCoreName: "aws.amazon.com/neuroncore"
|
||||
amd:
|
||||
resourceCountName: "amd.com/gpu"
|
||||
vnpus:
|
||||
- chipName: 910A
|
||||
commonWord: Ascend910A
|
||||
resourceName: huawei.com/Ascend910A
|
||||
resourceMemoryName: huawei.com/Ascend910A-memory
|
||||
memoryAllocatable: 32768
|
||||
memoryCapacity: 32768
|
||||
memoryFactor: 1
|
||||
aiCore: 30
|
||||
templates:
|
||||
- name: vir02
|
||||
memory: 2184
|
||||
aiCore: 2
|
||||
- name: vir04
|
||||
memory: 4369
|
||||
aiCore: 4
|
||||
- name: vir08
|
||||
memory: 8738
|
||||
aiCore: 8
|
||||
- name: vir16
|
||||
memory: 17476
|
||||
aiCore: 16
|
||||
- chipName: 910B2
|
||||
commonWord: Ascend910B2
|
||||
resourceName: huawei.com/Ascend910B2
|
||||
resourceMemoryName: huawei.com/Ascend910B2-memory
|
||||
memoryAllocatable: 65536
|
||||
memoryCapacity: 65536
|
||||
memoryFactor: 1
|
||||
aiCore: 24
|
||||
aiCPU: 6
|
||||
templates:
|
||||
- name: vir03_1c_8g
|
||||
memory: 8192
|
||||
aiCore: 3
|
||||
aiCPU: 1
|
||||
- name: vir06_1c_16g
|
||||
memory: 16384
|
||||
aiCore: 6
|
||||
aiCPU: 1
|
||||
- name: vir12_3c_32g
|
||||
memory: 32768
|
||||
aiCore: 12
|
||||
aiCPU: 3
|
||||
- chipName: 910B3
|
||||
commonWord: Ascend910B3
|
||||
resourceName: huawei.com/Ascend910B3
|
||||
resourceMemoryName: huawei.com/Ascend910B3-memory
|
||||
memoryAllocatable: 65536
|
||||
memoryCapacity: 65536
|
||||
memoryFactor: 1
|
||||
aiCore: 20
|
||||
aiCPU: 7
|
||||
templates:
|
||||
- name: vir05_1c_16g
|
||||
memory: 16384
|
||||
aiCore: 5
|
||||
aiCPU: 1
|
||||
- name: vir10_3c_32g
|
||||
memory: 32768
|
||||
aiCore: 10
|
||||
aiCPU: 3
|
||||
- chipName: 910B4-1
|
||||
commonWord: Ascend910B4-1
|
||||
resourceName: huawei.com/Ascend910B4-1
|
||||
resourceMemoryName: huawei.com/Ascend910B4-1-memory
|
||||
memoryAllocatable: 65536
|
||||
memoryCapacity: 65536
|
||||
memoryFactor: 1
|
||||
aiCore: 20
|
||||
aiCPU: 7
|
||||
templates:
|
||||
# NOTE: Names of vnpu template for 910B4-1 are fixed by Ascend runtime and must not be changed.
|
||||
# The memory is used for scheduling so the correct values must be set.
|
||||
# Template vir05_1c_8g actually provides 16GB memory,
|
||||
- name: vir05_1c_8g
|
||||
memory: 16384
|
||||
aiCore: 5
|
||||
aiCPU: 1
|
||||
# Template vir10_3c_16g actually provides 32GB memory
|
||||
- name: vir10_3c_16g
|
||||
memory: 32768
|
||||
aiCore: 10
|
||||
aiCPU: 3
|
||||
- chipName: 910B4
|
||||
commonWord: Ascend910B4
|
||||
resourceName: huawei.com/Ascend910B4
|
||||
resourceMemoryName: huawei.com/Ascend910B4-memory
|
||||
memoryAllocatable: 32768
|
||||
memoryCapacity: 32768
|
||||
memoryFactor: 1
|
||||
aiCore: 20
|
||||
aiCPU: 7
|
||||
templates:
|
||||
- name: vir05_1c_8g
|
||||
memory: 8192
|
||||
aiCore: 5
|
||||
aiCPU: 1
|
||||
- name: vir10_3c_16g
|
||||
memory: 16384
|
||||
aiCore: 10
|
||||
aiCPU: 3
|
||||
- chipName: 310P3
|
||||
commonWord: Ascend310P
|
||||
resourceName: huawei.com/Ascend310P
|
||||
resourceMemoryName: huawei.com/Ascend310P-memory
|
||||
memoryAllocatable: 21527
|
||||
memoryCapacity: 24576
|
||||
memoryFactor: 1
|
||||
aiCore: 8
|
||||
aiCPU: 7
|
||||
templates:
|
||||
- name: vir01
|
||||
memory: 3072
|
||||
aiCore: 1
|
||||
aiCPU: 1
|
||||
- name: vir02
|
||||
memory: 6144
|
||||
aiCore: 2
|
||||
aiCPU: 2
|
||||
- name: vir04
|
||||
memory: 12288
|
||||
aiCore: 4
|
||||
aiCPU: 4
|
||||
{{ end }}
|
||||
|
|
@ -0,0 +1,21 @@
|
|||
{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRole
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
annotations:
|
||||
"helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade
|
||||
"helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
rules:
|
||||
- apiGroups:
|
||||
- admissionregistration.k8s.io
|
||||
resources:
|
||||
#- validatingwebhookconfigurations
|
||||
- mutatingwebhookconfigurations
|
||||
verbs:
|
||||
- get
|
||||
- update
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,20 @@
|
|||
{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
annotations:
|
||||
"helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade
|
||||
"helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: ClusterRole
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,68 @@
|
|||
{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}}
|
||||
apiVersion: batch/v1
|
||||
kind: Job
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission-create
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
annotations:
|
||||
"helm.sh/hook": pre-install,pre-upgrade
|
||||
"helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
spec:
|
||||
{{- if .Capabilities.APIVersions.Has "batch/v1alpha1" }}
|
||||
# Alpha feature since k8s 1.12
|
||||
ttlSecondsAfterFinished: 0
|
||||
{{- end }}
|
||||
template:
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission-create
|
||||
{{- if .Values.scheduler.patch.podAnnotations }}
|
||||
annotations: {{ toYaml .Values.scheduler.patch.podAnnotations | nindent 8 }}
|
||||
{{- end }}
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 8 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
hami.io/webhook: ignore
|
||||
spec:
|
||||
{{- if .Values.scheduler.patch.priorityClassName }}
|
||||
priorityClassName: {{ .Values.scheduler.patch.priorityClassName }}
|
||||
{{- end }}
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
{{- include "hami.scheduler.patch.new.imagePullSecrets" . | nindent 6 }}
|
||||
{{- else }}
|
||||
{{- include "hami.scheduler.patch.imagePullSecrets" . | nindent 6 }}
|
||||
{{- end }}
|
||||
containers:
|
||||
- name: create
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
image: {{ include "hami.scheduler.patch.new.image" . }}
|
||||
imagePullPolicy: {{ .Values.scheduler.patch.imageNew.pullPolicy }}
|
||||
{{- else }}
|
||||
image: {{ include "hami.scheduler.patch.image" . }}
|
||||
imagePullPolicy: {{ .Values.scheduler.patch.image.pullPolicy }}
|
||||
{{- end }}
|
||||
args:
|
||||
- create
|
||||
- --cert-name=tls.crt
|
||||
- --key-name=tls.key
|
||||
{{- if .Values.scheduler.admissionWebhook.customURL.enabled }}
|
||||
- --host={{ printf "%s.%s.svc,127.0.0.1,%s" (include "hami-vgpu.scheduler" .) (include "hami-vgpu.namespace" .) .Values.scheduler.admissionWebhook.customURL.host}}
|
||||
{{- else }}
|
||||
- --host={{ printf "%s.%s.svc,127.0.0.1" (include "hami-vgpu.scheduler" .) (include "hami-vgpu.namespace" .) }}
|
||||
{{- end }}
|
||||
- --namespace={{ include "hami-vgpu.namespace" . }}
|
||||
- --secret-name={{ include "hami-vgpu.scheduler.tls" . }}
|
||||
restartPolicy: OnFailure
|
||||
serviceAccountName: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
{{- if .Values.scheduler.patch.nodeSelector }}
|
||||
nodeSelector: {{ toYaml .Values.scheduler.patch.nodeSelector | nindent 8 }}
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.patch.tolerations }}
|
||||
tolerations: {{ toYaml .Values.scheduler.patch.tolerations | nindent 8 }}
|
||||
{{- end }}
|
||||
securityContext:
|
||||
runAsNonRoot: true
|
||||
runAsUser: {{ .Values.scheduler.patch.runAsUser }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,63 @@
|
|||
{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}}
|
||||
apiVersion: batch/v1
|
||||
kind: Job
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission-patch
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
annotations:
|
||||
"helm.sh/hook": post-install,post-upgrade
|
||||
"helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
spec:
|
||||
{{- if .Capabilities.APIVersions.Has "batch/v1alpha1" }}
|
||||
# Alpha feature since k8s 1.12
|
||||
ttlSecondsAfterFinished: 0
|
||||
{{- end }}
|
||||
template:
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission-patch
|
||||
{{- if .Values.scheduler.patch.podAnnotations }}
|
||||
annotations: {{ toYaml .Values.scheduler.patch.podAnnotations | nindent 8 }}
|
||||
{{- end }}
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 8 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
hami.io/webhook: ignore
|
||||
spec:
|
||||
{{- if .Values.scheduler.patch.priorityClassName }}
|
||||
priorityClassName: {{ .Values.scheduler.patch.priorityClassName }}
|
||||
{{- end }}
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
{{- include "hami.scheduler.patch.new.imagePullSecrets" . | nindent 6 }}
|
||||
{{- else }}
|
||||
{{- include "hami.scheduler.patch.imagePullSecrets" . | nindent 6 }}
|
||||
{{- end }}
|
||||
containers:
|
||||
- name: patch
|
||||
{{- if ge (regexReplaceAll "[^0-9]" .Capabilities.KubeVersion.Minor "" | int) 22 }}
|
||||
image: {{ include "hami.scheduler.patch.new.image" . }}
|
||||
imagePullPolicy: {{ .Values.scheduler.patch.imageNew.pullPolicy }}
|
||||
{{- else }}
|
||||
image: {{ include "hami.scheduler.patch.image" . }}
|
||||
imagePullPolicy: {{ .Values.scheduler.patch.image.pullPolicy }}
|
||||
{{- end }}
|
||||
args:
|
||||
- patch
|
||||
- --webhook-name={{ include "hami-vgpu.scheduler.webhook" . }}
|
||||
- --namespace={{ include "hami-vgpu.namespace" . }}
|
||||
- --patch-validating=false
|
||||
- --secret-name={{ include "hami-vgpu.scheduler.tls" . }}
|
||||
restartPolicy: OnFailure
|
||||
serviceAccountName: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
{{- if .Values.scheduler.patch.nodeSelector }}
|
||||
nodeSelector: {{ toYaml .Values.scheduler.patch.nodeSelector | nindent 8 }}
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.patch.tolerations }}
|
||||
tolerations: {{ toYaml .Values.scheduler.patch.tolerations | nindent 8 }}
|
||||
{{- end }}
|
||||
securityContext:
|
||||
runAsNonRoot: true
|
||||
runAsUser: {{ .Values.scheduler.patch.runAsUser }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,21 @@
|
|||
{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: Role
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
annotations:
|
||||
"helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade
|
||||
"helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
rules:
|
||||
- apiGroups:
|
||||
- ""
|
||||
resources:
|
||||
- secrets
|
||||
verbs:
|
||||
- get
|
||||
- create
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,21 @@
|
|||
{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) -}}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: RoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
annotations:
|
||||
"helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade
|
||||
"helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: Role
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,13 @@
|
|||
{{- if and .Values.scheduler.admissionWebhook.enabled (.Values.scheduler.patch.enabled) (not .Values.scheduler.certManager.enabled) }}
|
||||
apiVersion: v1
|
||||
kind: ServiceAccount
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.fullname" . }}-admission
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
annotations:
|
||||
"helm.sh/hook": pre-install,pre-upgrade,post-install,post-upgrade
|
||||
"helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
|
||||
labels:
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
app.kubernetes.io/component: admission-webhook
|
||||
{{- end }}
|
||||
|
|
@ -0,0 +1,12 @@
|
|||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: Role
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-scheduler"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
rules:
|
||||
- apiGroups: ["coordination.k8s.io"]
|
||||
resources: ["leases"]
|
||||
verbs: ["create", "list", "watch", "get", "update", "patch"]
|
||||
|
|
@ -0,0 +1,31 @@
|
|||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: RoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-scheduler"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: Role
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
---
|
||||
{{- if .Values.mockDevicePlugin.enabled }}
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRoleBinding
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.mock-device-plugin" . }}
|
||||
roleRef:
|
||||
apiGroup: rbac.authorization.k8s.io
|
||||
kind: ClusterRole
|
||||
name: {{ include "hami-vgpu.mock-device-plugin" . }}
|
||||
subjects:
|
||||
- kind: ServiceAccount
|
||||
name: {{ include "hami-vgpu.mock-device-plugin" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,34 @@
|
|||
apiVersion: v1
|
||||
kind: Service
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
{{- if .Values.scheduler.service.labels }}
|
||||
{{ toYaml .Values.scheduler.service.labels | indent 4 }}
|
||||
{{- end }}
|
||||
{{- if .Values.scheduler.service.annotations }}
|
||||
annotations: {{ toYaml .Values.scheduler.service.annotations | nindent 4 }}
|
||||
{{- end }}
|
||||
spec:
|
||||
type: {{ .Values.scheduler.service.type | default "NodePort" }} # Default type is NodePort
|
||||
ports:
|
||||
- name: http
|
||||
port: {{ .Values.scheduler.service.httpPort | default 443 }} # Default HTTP port is 443
|
||||
targetPort: {{ .Values.scheduler.service.httpTargetPort | default 443 }}
|
||||
{{- if eq (.Values.scheduler.service.type | default "NodePort") "NodePort" }} # If type is NodePort, set nodePort
|
||||
nodePort: {{ .Values.scheduler.service.schedulerPort | default 31998 }}
|
||||
{{- end }}
|
||||
protocol: TCP
|
||||
- name: monitor
|
||||
port: {{ .Values.scheduler.service.monitorPort | default 31993 }} # Default monitoring port is 31993
|
||||
targetPort: {{ .Values.scheduler.service.monitorTargetPort | default 9395 }}
|
||||
{{- if eq (.Values.scheduler.service.type | default "NodePort") "NodePort" }} # If type is NodePort, set nodePort
|
||||
nodePort: {{ .Values.scheduler.service.monitorPort | default 31993 }}
|
||||
{{- end }}
|
||||
protocol: TCP
|
||||
selector:
|
||||
app.kubernetes.io/component: hami-scheduler
|
||||
{{- include "hami-vgpu.selectorLabels" . | nindent 4 }}
|
||||
|
|
@ -0,0 +1,16 @@
|
|||
apiVersion: v1
|
||||
kind: ServiceAccount
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
labels:
|
||||
app.kubernetes.io/component: "hami-scheduler"
|
||||
{{- include "hami-vgpu.labels" . | nindent 4 }}
|
||||
---
|
||||
{{- if .Values.mockDevicePlugin.enabled }}
|
||||
apiVersion: v1
|
||||
kind: ServiceAccount
|
||||
metadata:
|
||||
name: {{ include "hami-vgpu.mock-device-plugin" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
{{- end -}}
|
||||
|
|
@ -0,0 +1,57 @@
|
|||
{{- if .Values.scheduler.admissionWebhook.enabled -}}
|
||||
apiVersion: admissionregistration.k8s.io/v1
|
||||
kind: MutatingWebhookConfiguration
|
||||
metadata:
|
||||
{{- if .Values.scheduler.certManager.enabled }}
|
||||
annotations:
|
||||
cert-manager.io/inject-ca-from: {{ include "hami-vgpu.namespace" . }}/{{ include "hami-vgpu.scheduler" . }}-serving-cert
|
||||
{{- end }}
|
||||
name: {{ include "hami-vgpu.scheduler.webhook" . }}
|
||||
webhooks:
|
||||
- admissionReviewVersions:
|
||||
- v1beta1
|
||||
clientConfig:
|
||||
{{- if .Values.scheduler.admissionWebhook.customURL.enabled }}
|
||||
url: https://{{ .Values.scheduler.admissionWebhook.customURL.host}}:{{.Values.scheduler.admissionWebhook.customURL.port}}{{.Values.scheduler.admissionWebhook.customURL.path}}
|
||||
{{- else }}
|
||||
service:
|
||||
name: {{ include "hami-vgpu.scheduler" . }}
|
||||
namespace: {{ include "hami-vgpu.namespace" . }}
|
||||
path: /webhook
|
||||
port: {{ .Values.scheduler.service.httpPort }}
|
||||
{{- end }}
|
||||
failurePolicy: {{ .Values.scheduler.admissionWebhook.failurePolicy }}
|
||||
matchPolicy: Equivalent
|
||||
name: vgpu.hami.io
|
||||
namespaceSelector:
|
||||
matchExpressions:
|
||||
- key: hami.io/webhook
|
||||
operator: NotIn
|
||||
values:
|
||||
- ignore
|
||||
{{- if .Values.scheduler.admissionWebhook.whitelistNamespaces }}
|
||||
- key: kubernetes.io/metadata.name
|
||||
operator: NotIn
|
||||
values:
|
||||
{{- toYaml .Values.scheduler.admissionWebhook.whitelistNamespaces | nindent 10 }}
|
||||
{{- end }}
|
||||
objectSelector:
|
||||
matchExpressions:
|
||||
- key: hami.io/webhook
|
||||
operator: NotIn
|
||||
values:
|
||||
- ignore
|
||||
reinvocationPolicy: {{ .Values.scheduler.admissionWebhook.reinvocationPolicy }}
|
||||
rules:
|
||||
- apiGroups:
|
||||
- ""
|
||||
apiVersions:
|
||||
- v1
|
||||
operations:
|
||||
- CREATE
|
||||
resources:
|
||||
- pods
|
||||
scope: '*'
|
||||
sideEffects: None
|
||||
timeoutSeconds: 10
|
||||
{{- end }}
|
||||
455
packages/system/hami/charts/hami/values.yaml
Normal file
455
packages/system/hami/charts/hami/values.yaml
Normal file
|
|
@ -0,0 +1,455 @@
|
|||
## @section Global configuration
|
||||
## @param global.imageRegistry Global Docker image registry
|
||||
## @param global.imagePullSecrets Global Docker image pull secrets
|
||||
global:
|
||||
## @param global.imageRegistry Global Docker image registry
|
||||
imageRegistry: ""
|
||||
## E.g.
|
||||
## imagePullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
## @param global.imagePullSecrets Global Docker image pull secrets
|
||||
imagePullSecrets: []
|
||||
imageTag: "v2.8.1"
|
||||
gpuHookPath: /usr/local
|
||||
labels: {}
|
||||
annotations: {}
|
||||
managedNodeSelectorEnable: false
|
||||
managedNodeSelector:
|
||||
usage: "gpu"
|
||||
|
||||
nameOverride: ""
|
||||
fullnameOverride: ""
|
||||
namespaceOverride: ""
|
||||
|
||||
#Nvidia GPU Parameters
|
||||
resourceName: "nvidia.com/gpu"
|
||||
resourceMem: "nvidia.com/gpumem"
|
||||
resourceMemPercentage: "nvidia.com/gpumem-percentage"
|
||||
resourceCores: "nvidia.com/gpucores"
|
||||
resourcePriority: "nvidia.com/priority"
|
||||
|
||||
#MLU Parameters
|
||||
mluResourceName: "cambricon.com/vmlu"
|
||||
mluResourceMem: "cambricon.com/mlu.smlu.vmemory"
|
||||
mluResourceCores: "cambricon.com/mlu.smlu.vcore"
|
||||
|
||||
#Hygon DCU Parameters
|
||||
dcuResourceName: "hygon.com/dcunum"
|
||||
dcuResourceMem: "hygon.com/dcumem"
|
||||
dcuResourceCores: "hygon.com/dcucores"
|
||||
|
||||
#Metax sGPU Parameters
|
||||
metaxResourceName: "metax-tech.com/sgpu"
|
||||
metaxResourceCore: "metax-tech.com/vcore"
|
||||
metaxResourceMem: "metax-tech.com/vmemory"
|
||||
metaxsGPUTopologyAware: "false"
|
||||
|
||||
#Enflame VGCU Parameters
|
||||
enflameResourceNameVGCU: "enflame.com/vgcu"
|
||||
enflameResourceNameVGCUPercentage: "enflame.com/vgcu-percentage"
|
||||
|
||||
#Kunlun XPU Parameters
|
||||
kunlunResourceName: "kunlunxin.com/xpu"
|
||||
kunlunResourceVCountName: "kunlunxin.com/vxpu"
|
||||
kunlunResourceVMemoryName: "kunlunxin.com/vxpu-memory"
|
||||
|
||||
schedulerName: "hami-scheduler"
|
||||
|
||||
scheduler:
|
||||
# @param nodeName defines the node name and the nvidia-vgpu-scheduler-scheduler will schedule to the node.
|
||||
# if we install the nvidia-vgpu-scheduler-scheduler as default scheduler, we need to remove the k8s default
|
||||
# scheduler pod from the cluster first, we must specify node name to skip the schedule workflow.
|
||||
nodeName: ""
|
||||
#nodeLabelSelector:
|
||||
# "gpu": "on"
|
||||
overwriteEnv: "false"
|
||||
defaultSchedulerPolicy:
|
||||
nodeSchedulerPolicy: binpack
|
||||
gpuSchedulerPolicy: spread
|
||||
metricsBindAddress: ":9395"
|
||||
# If set to false, When Pod.Spec.SchedulerName equals to the const DefaultSchedulerName in k8s.io/api/core/v1 package, webhook will not overwrite it
|
||||
forceOverwriteDefaultScheduler: true
|
||||
livenessProbe: false
|
||||
leaderElect: true
|
||||
# when leaderElect is true, replicas is available, otherwise replicas is 1.
|
||||
replicas: 1
|
||||
kubeScheduler:
|
||||
# @param enabled indicate whether to run kube-scheduler container in the scheduler pod, it's true by default.
|
||||
enabled: true
|
||||
## @param image.registry kube scheduler image registry
|
||||
## @param image.repository kube scheduler image repository
|
||||
## @param image.tag kube scheduler image tag (immutable tags are recommended)
|
||||
## @param image.pullPolicy kube scheduler image pull policy
|
||||
## @param image.pullSecrets Specify docker-registry secret names as an array
|
||||
image:
|
||||
registry: "registry.cn-hangzhou.aliyuncs.com"
|
||||
repository: "google_containers/kube-scheduler"
|
||||
tag: ""
|
||||
## Specify a imagePullPolicy
|
||||
## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
|
||||
## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images
|
||||
##
|
||||
pullPolicy: IfNotPresent
|
||||
## Optionally specify an array of imagePullSecrets.
|
||||
## Secrets must be manually created in the namespace.
|
||||
## Example:
|
||||
## pullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
##
|
||||
pullSecrets: []
|
||||
resources: {}
|
||||
# If you do want to specify resources, uncomment the following lines, adjust them as necessary.
|
||||
# and remove the curly braces after 'resources:'.
|
||||
# limits:
|
||||
# cpu: 1000m
|
||||
# memory: 1000Mi
|
||||
# requests:
|
||||
# cpu: 100m
|
||||
# memory: 100Mi
|
||||
extraNewArgs:
|
||||
- --config=/config/config.yaml
|
||||
- -v=4
|
||||
extraArgs:
|
||||
- --policy-config-file=/config/config.json
|
||||
- -v=4
|
||||
extender:
|
||||
## @param image.registry scheduler extender image registry
|
||||
## @param image.repository scheduler extender image repository
|
||||
## @param image.tag scheduler extender image tag (immutable tags are recommended)
|
||||
## @param image.pullPolicy scheduler extender image pull policy
|
||||
## @param image.pullSecrets Specify docker-registry secret names as an array
|
||||
image:
|
||||
registry: "docker.io"
|
||||
repository: "projecthami/hami"
|
||||
tag: ""
|
||||
## Specify a imagePullPolicy
|
||||
## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
|
||||
## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images
|
||||
##
|
||||
pullPolicy: IfNotPresent
|
||||
## Optionally specify an array of imagePullSecrets.
|
||||
## Secrets must be manually created in the namespace.
|
||||
## Example:
|
||||
## pullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
##
|
||||
pullSecrets: []
|
||||
resources: {}
|
||||
# If you do want to specify resources, uncomment the following lines, adjust them as necessary,
|
||||
# and remove the curly braces after 'resources:'.
|
||||
# limits:
|
||||
# cpu: 1000m
|
||||
# memory: 1000Mi
|
||||
# requests:
|
||||
# cpu: 100m
|
||||
# memory: 100Mi
|
||||
extraArgs:
|
||||
- --debug
|
||||
- -v=4
|
||||
nodeLockExpire: "5m"
|
||||
podAnnotations: {}
|
||||
tolerations: []
|
||||
#serviceAccountName: "hami-vgpu-scheduler-sa"
|
||||
admissionWebhook:
|
||||
# If set to false, the admission webhook is not installed and any pods that should use HAMi must be
|
||||
# configured to use the right 'schedulerName' and other device-specific configurations.
|
||||
enabled: true
|
||||
customURL:
|
||||
enabled: false
|
||||
# must be an endpoint using https.
|
||||
# should generate host certs here
|
||||
host: 127.0.0.1 # hostname or ip, can be your node'IP if you want to use https://<nodeIP>:<schedulerPort>/<path>
|
||||
port: 31998
|
||||
path: /webhook
|
||||
whitelistNamespaces:
|
||||
# Specify the namespaces that the webhook will not be applied to.
|
||||
# - default
|
||||
# - kube-system
|
||||
# - istio-system
|
||||
reinvocationPolicy: Never
|
||||
failurePolicy: Ignore
|
||||
## TLS Certificate Option 1: Use cert-manager to generate self-signed certificate.
|
||||
## If enabled, always takes precedence over options 2.
|
||||
certManager:
|
||||
enabled: false
|
||||
## TLS Certificate Option 2: Use kube-webhook-certgen to generate self-signed certificate.
|
||||
## If true and certManager.enabled is false, Helm will automatically create a self-signed cert and secret for you.
|
||||
patch:
|
||||
enabled: true
|
||||
## @param image.registry scheduler extender image registry
|
||||
## @param image.repository scheduler extender image repository
|
||||
## @param image.tag scheduler extender image tag (immutable tags are recommended)
|
||||
## @param image.pullPolicy scheduler extender image pull policy
|
||||
## @param image.pullSecrets Specify docker-registry secret names as an array
|
||||
image:
|
||||
registry: "docker.io"
|
||||
repository: "jettech/kube-webhook-certgen"
|
||||
tag: "v1.5.2"
|
||||
## Specify a imagePullPolicy
|
||||
## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
|
||||
## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images
|
||||
##
|
||||
pullPolicy: IfNotPresent
|
||||
## Optionally specify an array of imagePullSecrets.
|
||||
## Secrets must be manually created in the namespace.
|
||||
## Example:
|
||||
## pullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
##
|
||||
pullSecrets: []
|
||||
## @param image.registry scheduler extender image registry
|
||||
## @param image.repository scheduler extender image repository
|
||||
## @param image.tag scheduler extender image tag (immutable tags are recommended)
|
||||
## @param image.pullPolicy scheduler extender image pull policy
|
||||
## @param image.pullSecrets Specify docker-registry secret names as an array
|
||||
imageNew:
|
||||
registry: "docker.io"
|
||||
repository: "liangjw/kube-webhook-certgen"
|
||||
tag: "v1.1.1"
|
||||
## Specify a imagePullPolicy
|
||||
## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
|
||||
## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images
|
||||
##
|
||||
pullPolicy: IfNotPresent
|
||||
## Optionally specify an array of imagePullSecrets.
|
||||
## Secrets must be manually created in the namespace.
|
||||
## Example:
|
||||
## pullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
##
|
||||
pullSecrets: []
|
||||
priorityClassName: ""
|
||||
podAnnotations: {}
|
||||
nodeSelector: {}
|
||||
tolerations: []
|
||||
runAsUser: 2000
|
||||
service:
|
||||
type: NodePort # Default type is NodePort, can be changed to ClusterIP
|
||||
httpPort: 443 # HTTP port
|
||||
schedulerPort: 31998 # NodePort for HTTP
|
||||
monitorPort: 31993 # Monitoring port
|
||||
monitorTargetPort: 9395
|
||||
httpTargetPort: 443
|
||||
labels: {}
|
||||
annotations: {}
|
||||
|
||||
devicePlugin:
|
||||
enabled: true
|
||||
gpuOperatorToolkitReady:
|
||||
enabled: false
|
||||
hostPath: "/run/nvidia/validations"
|
||||
## @param image.registry devicePlugin image registry
|
||||
## @param image.repository devicePlugin image repository
|
||||
## @param image.tag devicePlugin image tag (immutable tags are recommended)
|
||||
## @param image.pullPolicy devicePlugin image pull policy
|
||||
## @param image.pullSecrets Specify docker-registry secret names as an array
|
||||
image:
|
||||
registry: "docker.io"
|
||||
repository: "projecthami/hami"
|
||||
tag: ""
|
||||
## Specify a imagePullPolicy
|
||||
## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
|
||||
## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images
|
||||
##
|
||||
pullPolicy: IfNotPresent
|
||||
## Optionally specify an array of imagePullSecrets.
|
||||
## Secrets must be manually created in the namespace.
|
||||
## Example:
|
||||
## pullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
##
|
||||
pullSecrets: []
|
||||
monitor:
|
||||
## @param image.registry monitor image registry
|
||||
## @param image.repository monitor image repository
|
||||
## @param image.tag monitor image tag (immutable tags are recommended)
|
||||
## @param image.pullPolicy monitor image pull policy
|
||||
## @param image.pullSecrets Specify docker-registry secret names as an array
|
||||
image:
|
||||
registry: "docker.io"
|
||||
repository: "projecthami/hami"
|
||||
tag: ""
|
||||
## Specify a imagePullPolicy
|
||||
## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
|
||||
## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images
|
||||
##
|
||||
pullPolicy: IfNotPresent
|
||||
## Optionally specify an array of imagePullSecrets.
|
||||
## Secrets must be manually created in the namespace.
|
||||
## Example:
|
||||
## pullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
##
|
||||
pullSecrets: []
|
||||
ctrPath: /usr/local/vgpu/containers
|
||||
resyncInterval: "5m"
|
||||
extraArgs:
|
||||
- -v=4
|
||||
extraEnvs: {}
|
||||
resources: {}
|
||||
# If you do want to specify resources, uncomment the following lines, adjust them as necessary.
|
||||
# and remove the curly braces after 'resources:'.
|
||||
# limits:
|
||||
# cpu: 1000m
|
||||
# memory: 1000Mi
|
||||
# requests:
|
||||
# cpu: 100m
|
||||
# memory: 100Mi
|
||||
|
||||
deviceSplitCount: 10
|
||||
deviceMemoryScaling: 1
|
||||
deviceCoreScaling: 1
|
||||
# Node configuration for device plugin, Priority: externalConfigName > config > default config
|
||||
nodeConfiguration:
|
||||
# If you want to use a custom config.json, you can set the content here.
|
||||
# If this is set, it will override the default config.json(An example is as follows).
|
||||
config: |
|
||||
{
|
||||
"nodeconfig": [
|
||||
{
|
||||
"name": "your-node-name",
|
||||
"operatingmode": "hami-core",
|
||||
"devicememoryscaling": 1,
|
||||
"devicesplitcount": 10,
|
||||
"migstrategy": "none",
|
||||
"filterdevices": {
|
||||
"uuid": [],
|
||||
"index": []
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
# If you want to use an existing ConfigMap, you can set the name here.
|
||||
# If this is set, the chart will not create the ConfigMap and will use the existing one.
|
||||
externalConfigName: ""
|
||||
# The runtime class name to be used by the device plugin, and added to the pod.spec.runtimeClassName of applications utilizing NVIDIA GPUs
|
||||
runtimeClassName: ""
|
||||
# Whether to create runtime class, name comes from runtimeClassName when it is set
|
||||
createRuntimeClass: false
|
||||
migStrategy: "none"
|
||||
disablecorelimit: "false"
|
||||
passDeviceSpecsEnabled: false
|
||||
deviceListStrategy: "envvar"
|
||||
nvidiaHookPath: null
|
||||
nvidiaDriverRoot: null
|
||||
gdrcopyEnabled: null
|
||||
gdsEnabled: null
|
||||
mofedEnabled: null
|
||||
|
||||
extraArgs:
|
||||
- -v=4
|
||||
extraEnvs: {}
|
||||
service:
|
||||
type: NodePort # Default type is NodePort, can be changed to ClusterIP
|
||||
httpPort: 31992
|
||||
labels: {}
|
||||
annotations: {}
|
||||
|
||||
pluginPath: /var/lib/kubelet/device-plugins
|
||||
libPath: /usr/local/vgpu
|
||||
|
||||
podAnnotations: {}
|
||||
nvidiaNodeSelector:
|
||||
gpu: "on"
|
||||
tolerations: []
|
||||
# The updateStrategy for DevicePlugin DaemonSet.
|
||||
# If you want to update the DaemonSet by manual, set type as "OnDelete".
|
||||
# We recommend use OnDelete update strategy because DevicePlugin pod restart will cause business pod restart, this behavior is destructive.
|
||||
# Otherwise, you can use RollingUpdate update strategy to rolling update DevicePlugin pod.
|
||||
updateStrategy:
|
||||
type: RollingUpdate
|
||||
rollingUpdate:
|
||||
maxUnavailable: 1
|
||||
|
||||
resources: {}
|
||||
# If you do want to specify resources, uncomment the following lines, adjust them as necessary.
|
||||
# and remove the curly braces after 'resources:'.
|
||||
# limits:
|
||||
# cpu: 1000m
|
||||
# memory: 1000Mi
|
||||
# requests:
|
||||
# cpu: 100m
|
||||
# memory: 100Mi
|
||||
|
||||
mockDevicePlugin:
|
||||
enabled: false
|
||||
image:
|
||||
registry: "docker.io"
|
||||
repository: "projecthami/mock-device-plugin"
|
||||
tag: "1.0.1"
|
||||
## Specify a imagePullPolicy
|
||||
## Defaults to 'Always' if image tag is 'latest', else set to 'IfNotPresent'
|
||||
## ref: https://kubernetes.io/docs/user-guide/images/#pre-pulling-images
|
||||
##
|
||||
pullPolicy: IfNotPresent
|
||||
## Optionally specify an array of imagePullSecrets.
|
||||
## Secrets must be manually created in the namespace.
|
||||
## Example:
|
||||
## pullSecrets:
|
||||
## - myRegistryKeySecretName
|
||||
##
|
||||
pullSecrets: []
|
||||
devices:
|
||||
amd:
|
||||
customresources:
|
||||
- amd.com/gpu
|
||||
- amd.com/gpu-memory
|
||||
awsneuron:
|
||||
customresources:
|
||||
- aws.amazon.com/neuron
|
||||
- aws.amazon.com/neuroncore
|
||||
kunlun:
|
||||
enabled: true
|
||||
customresources:
|
||||
- kunlunxin.com/xpu
|
||||
- kunlunxin.com/vxpu
|
||||
- kunlunxin.com/vxpu-memory
|
||||
enflame:
|
||||
enabled: true
|
||||
customresources:
|
||||
- enflame.com/vgcu
|
||||
- enflame.com/vgcu-percentage
|
||||
- enflame.com/gcu
|
||||
mthreads:
|
||||
enabled: true
|
||||
customresources:
|
||||
- mthreads.com/vgpu
|
||||
nvidia:
|
||||
gpuCorePolicy: default
|
||||
libCudaLogLevel: 1
|
||||
ascend:
|
||||
enabled: false
|
||||
image: ""
|
||||
imagePullPolicy: IfNotPresent
|
||||
extraArgs: []
|
||||
nodeSelector:
|
||||
ascend: "on"
|
||||
tolerations: []
|
||||
customresources:
|
||||
- huawei.com/Ascend910A
|
||||
- huawei.com/Ascend910A-memory
|
||||
- huawei.com/Ascend910B2
|
||||
- huawei.com/Ascend910B2-memory
|
||||
- huawei.com/Ascend910B3
|
||||
- huawei.com/Ascend910B3-memory
|
||||
- huawei.com/Ascend910B4
|
||||
- huawei.com/Ascend910B4-memory
|
||||
- huawei.com/Ascend910B4-1
|
||||
- huawei.com/Ascend910B4-1-memory
|
||||
- huawei.com/Ascend310P
|
||||
- huawei.com/Ascend310P-memory
|
||||
iluvatar:
|
||||
enabled: false
|
||||
customresources:
|
||||
- iluvatar.ai/BI-V100-vgpu
|
||||
- iluvatar.ai/BI-V100.vCore
|
||||
- iluvatar.ai/BI-V100.vMem
|
||||
- iluvatar.ai/BI-V150-vgpu
|
||||
- iluvatar.ai/BI-V150.vCore
|
||||
- iluvatar.ai/BI-V150.vMem
|
||||
- iluvatar.ai/MR-V100-vgpu
|
||||
- iluvatar.ai/MR-V100.vCore
|
||||
- iluvatar.ai/MR-V100.vMem
|
||||
- iluvatar.ai/MR-V50-vgpu
|
||||
- iluvatar.ai/MR-V50.vCore
|
||||
- iluvatar.ai/MR-V50.vMem
|
||||
17
packages/system/hami/values.yaml
Normal file
17
packages/system/hami/values.yaml
Normal file
|
|
@ -0,0 +1,17 @@
|
|||
hami:
|
||||
scheduler:
|
||||
kubeScheduler:
|
||||
image:
|
||||
registry: registry.k8s.io
|
||||
repository: kube-scheduler
|
||||
devicePlugin:
|
||||
runtimeClassName: nvidia
|
||||
updateStrategy:
|
||||
type: RollingUpdate
|
||||
# See https://github.com/Project-HAMi/HAMi/blob/v2.8.1/deployments/charts/hami/values.yaml
|
||||
# Each entry: {"name": "node-name", "operatingmode": "hami-core", "devicesplitcount": N, ...}
|
||||
nodeConfiguration:
|
||||
config: |
|
||||
{
|
||||
"nodeconfig": []
|
||||
}
|
||||
File diff suppressed because one or more lines are too long
Loading…
Add table
Add a link
Reference in a new issue