Add Helm chart for kubeflow trainer

Signed-off-by: Yi Chen <github@chenyicn.net>

Makefile

@@ -11,6 +11,21 @@ endif
 SHELL = /usr/bin/env bash -o pipefail
 .SHELLFLAGS = -ec
 
+REPO := github.com/kubeflow/trainer
+TRAINER_CHART_PATH := charts/trainer
+
+## Location to install binaries
+LOCALBIN ?= $(shell pwd)/bin
+
+## Versions
+HELM_VERSION ?= v3.15.3
+HELM_UNITTEST_VERSION ?= 0.5.1
+HELM_DOCS_VERSION ?= v1.14.2
+
+## Binaries
+HELM ?= $(LOCALBIN)/helm-$(HELM_VERSION)
+HELM_DOCS ?= $(LOCALBIN)/helm-docs-$(HELM_DOCS_VERSION)
+
 ##@ General
 
 # The help target prints out all targets with their descriptions organized
@@ -124,3 +139,50 @@ test-python-integration: ## Run Python integration test.
 	pip install -r ./cmd/initializer/dataset/requirements.txt
 
 	pytest ./test/integration/initializer
+
+##@ Helm
+
+.PHONY: helm-unittest
+helm-unittest: helm-unittest-plugin ## Run Helm chart unittests.
+	$(HELM) unittest $(TRAINER_CHART_PATH) --strict --file "tests/**/*_test.yaml"
+
+.PHONY: helm-lint
+helm-lint: ## Run Helm chart lint test.
+	docker run --rm --workdir /workspace --volume "$$(pwd):/workspace" quay.io/helmpack/chart-testing:latest ct lint --target-branch master --validate-maintainers=false
+
+.PHONY: helm-docs
+helm-docs: helm-docs-plugin ## Generates markdown documentation for helm charts from requirements and values files.
+	$(HELM_DOCS) --sort-values-order=file
+
+##@ Dependencies
+
+.PHONY: helm
+helm: $(HELM) ## Download helm locally if necessary.
+$(HELM): $(LOCALBIN)
+	$(call go-install-tool,$(HELM),helm.sh/helm/v3/cmd/helm,$(HELM_VERSION))
+
+.PHONY: helm-unittest-plugin
+helm-unittest-plugin: helm ## Download helm unittest plugin locally if necessary.
+	if [ -z "$(shell $(HELM) plugin list | grep unittest)" ]; then \
+		echo "Installing helm unittest plugin"; \
+		$(HELM) plugin install https://github.com/helm-unittest/helm-unittest.git --version $(HELM_UNITTEST_VERSION); \
+	fi
+
+.PHONY: helm-docs-plugin
+helm-docs-plugin: $(HELM_DOCS) ## Download helm-docs plugin locally if necessary.
+$(HELM_DOCS): $(LOCALBIN)
+	$(call go-install-tool,$(HELM_DOCS),github.com/norwoodj/helm-docs/cmd/helm-docs,$(HELM_DOCS_VERSION))
+
+# go-install-tool will 'go install' any package with custom target and name of binary, if it doesn't exist
+# $1 - target path with name of binary (ideally with version)
+# $2 - package url which can be installed
+# $3 - specific version of package
+define go-install-tool
+@[ -f $(1) ] || { \
+set -e; \
+package=$(2)@$(3) ;\
+echo "Downloading $${package}" ;\
+GOBIN=$(LOCALBIN) go install $${package} ;\
+mv "$$(echo "$(1)" | sed "s/-$(3)$$//")" $(1) ;\
+}
+endef

charts/trainer/.helmignore

@@ -0,0 +1,23 @@
+# Patterns to ignore when building packages.
+# This supports shell glob matching, relative path matching, and
+# negation (prefixed with !). Only one pattern per line.
+.DS_Store
+# Common VCS dirs
+.git/
+.gitignore
+.bzr/
+.bzrignore
+.hg/
+.hgignore
+.svn/
+# Common backup files
+*.swp
+*.bak
+*.tmp
+*.orig
+*~
+# Various IDEs
+.project
+.idea/
+*.tmproj
+.vscode/

charts/trainer/Chart.yaml

@@ -0,0 +1,37 @@
+#
+# Copyright 2024 The Kubeflow authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+apiVersion: v2
+
+name: trainer
+
+description: A Helm chart for deploying Kubeflow trainer on Kubernetes.
+
+version: 2.0.0
+
+appVersion: 2.0.0
+
+type: application
+
+keywords:
+- kubeflow trainer
+
+home: https://github.com/kubeflow/trainer
+
+maintainers:
+- name: ChenYi015
+  email: github@chenyicn.net
+  url: https://github.com/ChenYi015

charts/trainer/README.md

@@ -0,0 +1,105 @@
+# trainer
+
+![Version: 2.0.0](https://img.shields.io/badge/Version-2.0.0-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: 2.0.0](https://img.shields.io/badge/AppVersion-2.0.0-informational?style=flat-square)
+
+A Helm chart for deploying Kubeflow trainer on Kubernetes.
+
+**Homepage:** <https://github.com/kubeflow/trainer>
+
+## Introduction
+
+This chart bootstraps a [Kubernetes Trainer](https://github.com/kubeflow/trainer) deployment using the [Helm](https://helm.sh) package manager.
+
+## Prerequisites
+
+- Helm >= 3
+- Kubernetes >= 1.20
+
+## Usage
+
+### Add Helm Repo
+
+```bash
+helm repo add trainer https://kubeflow.github.io/trainer
+
+helm repo update
+```
+
+See [helm repo](https://helm.sh/docs/helm/helm_repo) for command documentation.
+
+### Install the chart
+
+```bash
+helm install [RELEASE_NAME] trainer/trainer
+```
+
+For example, if you want to create a release with name `trainer` in the `kubeflow-system` namespace:
+
+```shell
+helm install trainer trainer/trainer \
+    --namespace kubeflow-system \
+    --create-namespace
+```
+
+Note that by passing the `--create-namespace` flag to the `helm install` command, `helm` will create the release namespace if it does not exist.
+
+See [helm install](https://helm.sh/docs/helm/helm_install) for command documentation.
+
+### Upgrade the chart
+
+```shell
+helm upgrade [RELEASE_NAME] trainer/trainer [flags]
+```
+
+See [helm upgrade](https://helm.sh/docs/helm/helm_upgrade) for command documentation.
+
+### Uninstall the chart
+
+```shell
+helm uninstall [RELEASE_NAME]
+```
+
+This removes all the Kubernetes resources associated with the chart and deletes the release, except for the `crds`, those will have to be removed manually.
+
+See [helm uninstall](https://helm.sh/docs/helm/helm_uninstall) for command documentation.
+
+## Values
+
+| Key | Type | Default | Description |
+|-----|------|---------|-------------|
+| nameOverride | string | `""` | String to partially override release name. |
+| fullnameOverride | string | `""` | String to fully override release name. |
+| commonLabels | object | `{}` | Common labels to add to the resources. |
+| image.registry | string | `"docker.io"` | Image registry. |
+| image.repository | string | `"kubeflow/trainer-controller-controller"` | Image repository. |
+| image.tag | string | If not set, the chart appVersion will be used. | Image tag. |
+| image.pullPolicy | string | `"IfNotPresent"` | Image pull policy. |
+| image.pullSecrets | list | `[]` | Image pull secrets for private image registry. |
+| controller.replicas | int | `1` | Number of replicas of controller. |
+| controller.labels | object | `{}` | Extra labels for controller pods. |
+| controller.annotations | object | `{}` | Extra annotations for controller pods. |
+| controller.volumes | list | `[]` | Volumes for controller pods. |
+| controller.nodeSelector | object | `{}` | Node selector for controller pods. |
+| controller.affinity | object | `{}` | Affinity for controller pods. |
+| controller.tolerations | list | `[]` | List of node taints to tolerate for controller pods. |
+| controller.env | list | `[]` | Environment variables for controller containers. |
+| controller.envFrom | list | `[]` | Environment variable sources for controller containers. |
+| controller.volumeMounts | list | `[]` | Volume mounts for controller containers. |
+| controller.resources | object | `{}` | Pod resource requests and limits for controller containers. |
+| controller.securityContext | object | `{}` | Security context for controller containers. |
+| controller.serviceAccount.create | bool | `true` | Specifies whether to create a service account for the controller. |
+| controller.serviceAccount.name | string | `""` | Optional name for the controller service account. |
+| controller.serviceAccount.annotations | object | `{}` | Extra annotations for the controller service account. |
+| controller.serviceAccount.automountServiceAccountToken | bool | `true` | Auto-mount service account token to the controller pods. |
+| webhook.enable | bool | `true` | Specifies whether to enable webhook. |
+| webhook.failurePolicy | string | `"Fail"` | Specifies how unrecognized errors are handled. Available options are `Ignore` or `Fail`. |
+| runtime.preTraining.torchDistributed.enable | bool | `true` |  |
+| runtime.preTraining.torchDistributed.image.registry | string | `"docker.io"` |  |
+| runtime.preTraining.torchDistributed.image.repository | string | `"pytorch/pytorch"` |  |
+| runtime.preTraining.torchDistributed.image.tag | string | `"2.5.0-cuda12.4-cudnn9-runtime"` |  |
+
+## Maintainers
+
+| Name | Email | Url |
+| ---- | ------ | --- |
+| ChenYi015 | <github@chenyicn.net> | <https://github.com/ChenYi015> |

charts/trainer/README.md.gotmpl

@@ -0,0 +1,70 @@
+{{ template "chart.header" . }}
+
+{{ template "chart.deprecationWarning" . }}
+
+{{ template "chart.badgesSection" . }}
+
+{{ template "chart.description" . }}
+
+{{ template "chart.homepageLine" . }}
+
+## Introduction
+
+This chart bootstraps a [Kubernetes Trainer]({{template "chart.homepage" . }}) deployment using the [Helm](https://helm.sh) package manager.
+
+## Prerequisites
+
+- Helm >= 3
+- Kubernetes >= 1.20
+
+## Usage
+
+### Add Helm Repo
+
+```bash
+helm repo add trainer https://kubeflow.github.io/trainer
+
+helm repo update
+```
+
+See [helm repo](https://helm.sh/docs/helm/helm_repo) for command documentation.
+
+### Install the chart
+
+```bash
+helm install [RELEASE_NAME] trainer/trainer
+```
+
+For example, if you want to create a release with name `trainer` in the `kubeflow-system` namespace:
+
+```shell
+helm install trainer trainer/trainer \
+    --namespace kubeflow-system \
+    --create-namespace
+```
+
+Note that by passing the `--create-namespace` flag to the `helm install` command, `helm` will create the release namespace if it does not exist.
+
+See [helm install](https://helm.sh/docs/helm/helm_install) for command documentation.
+
+### Upgrade the chart
+
+```shell
+helm upgrade [RELEASE_NAME] trainer/trainer [flags]
+```
+
+See [helm upgrade](https://helm.sh/docs/helm/helm_upgrade) for command documentation.
+
+### Uninstall the chart
+
+```shell
+helm uninstall [RELEASE_NAME]
+```
+
+This removes all the Kubernetes resources associated with the chart and deletes the release, except for the `crds`, those will have to be removed manually.
+
+See [helm uninstall](https://helm.sh/docs/helm/helm_uninstall) for command documentation.
+
+{{ template "chart.valuesSection" . }}
+
+{{ template "chart.maintainersSection" . }}