Merge pull request #1 from castai/create-scraper

Initialize project

.github/workflows/build.yaml

@@ -0,0 +1,92 @@
+name: Build
+
+on:
+  push:
+    branches:
+      - main
+  release:
+    types:
+      - published
+  pull_request:
+    branches:
+      - main
+
+jobs:
+  build:
+    name: Build
+    runs-on: ubuntu-20.04
+    steps:
+
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Setup Go
+        uses: actions/setup-go@v5
+        with:
+          go-version: 1.22
+
+
+      - name: Get release tag
+        if: github.event_name == 'release'
+        run: echo "RELEASE_TAG=${GITHUB_REF#refs/*/}" >> $GITHUB_ENV
+
+      - name: Build Go binary amd64
+        run: go build -ldflags "-s -w -X main.GitCommit=$GITHUB_SHA -X main.GitRef=$GITHUB_REF -X main.Version=${RELEASE_TAG:-commit-$GITHUB_SHA}" -o bin/gpu-metrics-exporter-amd64 ./cmd/main.go
+        env:
+          GOOS: linux
+          GOARCH: amd64
+          CGO_ENABLED: 0
+
+      - name: Build Go binary arm64
+        run: go build -ldflags "-s -w -X main.GitCommit=$GITHUB_SHA -X main.GitRef=$GITHUB_REF -X main.Version=${RELEASE_TAG:-commit-$GITHUB_SHA}" -o bin/gpu-metrics-exporter-arm64 ./cmd/main.go
+        env:
+          GOOS: linux
+          GOARCH: arm64
+          CGO_ENABLED: 0
+
+      - name: Test
+        run: go test -race ./...
+
+      # - name: Set up QEMU
+      #   uses: docker/setup-qemu-action@v3
+
+      # - name: Set up Docker Buildx
+      #   uses: docker/setup-buildx-action@v3
+
+      # - name: Login to GitHub Container Registry
+      #   uses: docker/login-action@v2
+      #   with:
+      #     registry: ghcr.io
+      #     username: ${{ github.actor }}
+      #     password: ${{ secrets.GITHUB_TOKEN }}
+
+      # - name: Build and push PR
+      #   if: ${{ github.event_name == 'pull_request' }}
+      #   uses: docker/build-push-action@v5
+      #   with:
+      #     context: .
+      #     platforms: linux/arm64,linux/amd64
+      #     file: ./Dockerfile
+      #     push: true
+      #     tags: ghcr.io/castai/gpu-metrics-exporter/gpu-metrics-exporter:${{ github.sha }}
+
+      # - name: Build and push main
+      #   if: ${{ github.event_name != 'pull_request' && github.event_name != 'release' }}
+      #   uses: docker/build-push-action@v5
+      #   with:
+      #     context: .
+      #     platforms: linux/arm64,linux/amd64
+      #     file: ./Dockerfile
+      #     push: true
+      #     tags: ghcr.io/castai/egressd/egressd:${{ github.sha }}
+
+      # - name: Build and push release (egressd collector)
+      #   uses: docker/build-push-action@v5
+      #   with:
+      #     context: .
+      #     push: true
+      #     platforms: linux/arm64,linux/amd64
+      #     file: ./Dockerfile
+      #     tags: |
+      #       ghcr.io/castai/gpu-metrics-exporter/gpu-metrics-exporter:${{ env.RELEASE_TAG }}
+      #       ghcr.io/castai/gpu-metrics-exporter/gpu-metrics-exporter:latest

.github/workflows/golangci-lint.yaml

@@ -0,0 +1,18 @@
+name: golangci-lint
+on:
+  pull_request:
+permissions:
+  contents: read
+jobs:
+  golangci:
+    name: lint
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-go@v5
+        with:
+          go-version: 1.22.0
+      - name: golangci-lint
+        uses: golangci/golangci-lint-action@v4
+        with:
+          args: --timeout=5m

.gitignore

@@ -0,0 +1,3 @@
+bin
+coverage.txt
+.vscode

.golangci.yaml

@@ -0,0 +1,87 @@
+linters-settings:
+  golint:
+    min-confidence: 0
+  gomnd:
+    settings:
+      mnd:
+        # don't include the "operation" and "assign"
+        checks: [argument,case,condition,return]
+  govet:
+    # shadow is marked as experimental feature, skip it for now.
+    check-shadowing: false
+    settings:
+      printf:
+        funcs:
+          - (github.com/golangci/golangci-lint/pkg/logutils.Log).Infof
+          - (github.com/golangci/golangci-lint/pkg/logutils.Log).Warnf
+          - (github.com/golangci/golangci-lint/pkg/logutils.Log).Errorf
+          - (github.com/golangci/golangci-lint/pkg/logutils.Log).Fatalf
+  lll:
+    line-length: 200
+  maligned:
+    suggest-new: true
+  misspell:
+    locale: US
+  revive:
+    rules:
+      - name: redefines-builtin-id
+        disabled: true
+      - name: nested-structs
+        disabled: true
+  gci:
+    sections:
+      - standard
+      - default
+      - prefix(github.com/castai)
+linters:
+  disable-all: true
+  enable:
+    - dogsled
+    - errcheck
+    - gofmt
+    - revive
+    - goprintffuncname
+    - gosimple
+    - gosec
+    - govet
+    - ineffassign
+    - lll
+    - misspell
+    - nakedret
+    - exportloopref
+    - staticcheck
+    - stylecheck
+    - typecheck
+    - unconvert
+    - unused
+    - whitespace
+    - errname
+    - dupword
+    - gci
+    - containedctx
+    - durationcheck
+    - errorlint
+issues:
+  # Excluding configuration per-path, per-linter, per-text and per-source
+  exclude-rules:
+    - path: _test\.go
+      linters:
+        - gomnd
+        - bodyclose
+        - gosec
+    - linters:
+        - gosec
+      # Ignored gosec G107 rule because of many false positives. It states that `http.Get(url)` must not contain a variable.
+      text: G107
+    - linters:
+        - unparam
+      # ignoring error where function always receives same value - mostly in tests
+      text: "always receives"
+    - linters:
+        - revive
+      # tolerate code where errors are returned as if-return
+      text: "redundant if ...; err != nil check, just return error instead."
+run:
+  skip-dirs:
+    - gen
+    - e2e

.mockery.yaml

@@ -0,0 +1,9 @@
+with-expecter: true
+dir: 'mock/{{replace .InterfaceDirRelative "internal" "" 1}}'
+packages:
+  "github.com/castai/gpu-metrics-exporter/internal/exporter":
+    interfaces:
+      Exporter:
+      Scraper:
+      MetricMapper:
+      HttpClient:

Dockerfile

@@ -0,0 +1,4 @@
+FROM gcr.io/distroless/static-debian11:nonroot
+ARG TARGETARCH
+COPY bin/gpu-metrics-exporter-$TARGETARCH /usr/local/bin/gpu-metrics-exporter
+CMD ["gpu-metrics-exporter"]

Makefile

@@ -0,0 +1,38 @@
+.PHONY: lint
+lint: check-lint-dependencies
+	golangci-lint run ./...
+
+.PHONY: fix-lint-lint
+fix-lint: check-lint-dependencies
+	golangci-lint run ./... --fix
+
+.PHONY: build
+build:
+	GOOS=linux GOARCH=amd64 CGO_ENABLED=0 go build -ldflags "-s -w" -o bin/gpu-metrics-exporter-amd64 ./cmd/main.go
+	GOOS=linux GOARCH=arm64 CGO_ENABLED=0 go build -ldflags "-s -w" -o bin/gpu-metrics-exporter-arm64 ./cmd/main.go
+	docker buildx build --push --platform=linux/amd64,linux/arm64 -t  $(TAG) .
+
+.PHONY: generate
+generate:
+	go generate ./...
+
+SHELL := /bin/bash
+.PHONY: run
+run:
+	go run ./cmd/main.go
+
+.PHONY: test
+test:
+	go test ./... -race -coverprofile=coverage.txt -covermode=atomic
+
+.PHONY: gen-proto
+gen-proto: check-proto-dependencies
+	protoc pb/metrics.proto --go_out=paths=source_relative:.
+
+.PHONY: check-lint-dependencies
+check-lint-dependencies:
+	@which golangci-lint > /dev/null || (echo "golangci-lint not found, please install it" && exit 1)
+
+.PHONY: check-proto-dependencies
+check-proto-dependencies:
+	@which protoc > /dev/null || (echo "protoc not found, please install it" && exit 1)

cmd/main.go

@@ -0,0 +1,128 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"net/http"
+	"os"
+	"os/signal"
+	"syscall"
+	"time"
+
+	"github.com/sirupsen/logrus"
+	"k8s.io/apimachinery/pkg/labels"
+	"k8s.io/apimachinery/pkg/selection"
+	"k8s.io/client-go/kubernetes"
+	"k8s.io/client-go/tools/clientcmd"
+	"k8s.io/client-go/util/flowcontrol"
+
+	"github.com/castai/gpu-metrics-exporter/internal/config"
+	"github.com/castai/gpu-metrics-exporter/internal/exporter"
+	"github.com/castai/gpu-metrics-exporter/internal/server"
+)
+
+func main() {
+	log := logrus.New()
+
+	cfg, err := config.GetFromEnvironment()
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	logLevel, err := logrus.ParseLevel(cfg.LogLevel)
+	if err != nil {
+		log.Fatal(err)
+	}
+	log.SetLevel(logLevel)
+
+	if err := run(cfg, log); err != nil && !errors.Is(err, context.Canceled) {
+		log.Fatal(err)
+	}
+}
+
+func run(cfg *config.Config, log logrus.FieldLogger) error {
+	mux := server.NewServerMux(log)
+
+	srv := &http.Server{
+		Addr:              fmt.Sprintf(":%d", cfg.HTTPListenPort),
+		Handler:           mux,
+		ReadHeaderTimeout: 3 * time.Second,
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	go func() {
+		stopper := make(chan os.Signal, 1)
+		signal.Notify(stopper, os.Interrupt, syscall.SIGTERM, syscall.SIGINT)
+		<-stopper
+
+		ctx, shutdownCancel := context.WithTimeout(context.Background(), 10*time.Second)
+		defer shutdownCancel()
+		if err := srv.Shutdown(ctx); err != nil {
+			log.Errorf("http server shutdown: %v", err)
+		}
+
+		cancel()
+	}()
+
+	clientset, err := newKubernetesClientset(cfg)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	labelSelector, err := selectorFromMap(cfg.DCGMLabels)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	scraper := exporter.NewScraper(&http.Client{}, log)
+	mapper := exporter.NewMapper()
+	ex := exporter.NewExporter(exporter.Config{
+		ExportInterval:   cfg.ExportInterval,
+		Selector:         labelSelector.String(),
+		DCGMExporterPort: cfg.DCGMPort,
+		DCGMExporterPath: cfg.DCGMMetricsEndpoint,
+		Enabled:          true,
+	}, clientset, log, scraper, mapper)
+
+	go func() {
+		if err := ex.Start(ctx); err != nil && !errors.Is(err, context.Canceled) {
+			log.Errorf("exporter stopped with error %v", err)
+			cancel()
+		}
+	}()
+
+	return srv.ListenAndServe()
+}
+
+func newKubernetesClientset(cfg *config.Config) (*kubernetes.Clientset, error) {
+	config, err := clientcmd.BuildConfigFromFlags("", cfg.KubeConfigPath)
+	if err != nil {
+		return nil, err
+	}
+	config.RateLimiter = flowcontrol.NewTokenBucketRateLimiter(float32(10), 25)
+
+	clientset, err := kubernetes.NewForConfig(config)
+	if err != nil {
+		return nil, err
+	}
+
+	return clientset, nil
+}
+
+func selectorFromMap(labelMap map[string]string) (labels.Selector, error) {
+	selector := labels.NewSelector()
+	var requirements labels.Requirements
+
+	for label, value := range labelMap {
+		req, err := labels.NewRequirement(label, selection.Equals, []string{value})
+		if err != nil {
+			return nil, err
+		}
+		requirements = append(requirements, *req)
+	}
+
+	return selector.Add(requirements...), nil
+}

gen_mockery.go

@@ -0,0 +1,2 @@
+//go:generate go run github.com/vektra/mockery/[email protected] --all
+package mockery