feat: otel instrumentation of wundernode and server (#1004)

* feat: otel instrumentation

* chore: add sdk interface

* chore: upgrade to go 1.19.0 due to otel dep

* chore: zero overhead when disabling otel

* chore: add example, improve error reporting

* chore: ensure same default

* chore: add e2e tests

* chore: address feedback

* chore: limit concurrency for big machines

* chore: update vitest

* chore: upgrade to go 1.20 and docker compose 3.8

* chore: run with limited concurrency, improves performance and stability


* chore: fix port race in tests

Author: StarpTech

.github/actions/go/action.yaml

@@ -3,7 +3,7 @@ description: 'This action install go and cache modules and build dependencies'
 inputs:
   go-version:
     description: 'The go version to install'
-    default: '1.18'
+    default: '1.20'
     required: false
   working-directory:
     description: 'The working directory of your go module'

.github/workflows/engine-ci.yaml

@@ -73,7 +73,8 @@ jobs:
       - name: Linters
         uses: golangci/golangci-lint-action@v3
         with:
-          version: v1.50
+          # keep in sync with local version
+          version: v1.52.2
           args: --timeout 2m
 
       - name: Run unit and integration tests

.github/workflows/packages-ci.yaml

@@ -78,7 +78,7 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Run testsuite
-        run: pnpm test:suite
+        run: pnpm test:suite:ci
 
       - name: Determine changed files
         id: changed

CONTRIBUTING.md

@@ -9,7 +9,7 @@ This guide assumes you have already installed the following software:
 - make (should be installed on all linux / IOS systems)
 - [protoc](https://grpc.io/docs/protoc-installation/) (3.15.8)
   - curl -LO "https://github.com/protocolbuffers/protobuf/releases/download/v21.5/protoc-21.5-linux-x86_64.zip"
-- golang `>= 1.18`
+- golang `>= 1.20`
 - Node.js [LTS](https://nodejs.org/en/about/releases/)
 - pnpm `7`
 

Dockerfile

@@ -1,4 +1,4 @@
-FROM golang:1.18 as builder
+FROM golang:1.20 as builder
 
 ARG COMMIT
 ARG VERSION
@@ -12,7 +12,7 @@ RUN go mod download
 
 RUN make build
 
-from golang:1.18-alpine as runner
+from golang:1.20-alpine as runner
 
 WORKDIR /usr/src/app
 

Makefile

@@ -50,7 +50,7 @@ codegen: install-proto codegen-go
 	pnpm codegen
 
 build: codegen
-	cd cmd/wunderctl && go build -o ../../wunderctl -ldflags "-X 'main.commit=$(shell git rev-parse --short HEAD)' -X 'main.builtBy=dev' -X 'main.version=dev' -X 'main.date=$(shell date)'" -trimpath
+	cd cmd/wunderctl && CGO_ENABLED=0 go build -o ../../wunderctl -ldflags "-X 'main.commit=$(shell git rev-parse --short HEAD)' -X 'main.builtBy=dev' -X 'main.version=dev' -X 'main.date=$(shell date)'" -trimpath
 
 # This command builds the wunderctl binary and copies it into the nodejs wunderctl wrapper
 wunderctl: build

README.md

@@ -244,6 +244,7 @@ The best way to try them out is by cloning the mono-repo.
 
 - [Use Turborepo to speed up builds in monorepos.](https://github.com/wundergraph/cloud-starter-turbo)
 - [Manage multiple WunderGraph Cloud Projects in a single repository](https://github.com/wundergraph/cloud-multi-projects-demo)
+- [Built-in OpenTelemetry instrumentation](examples/openapi-mocking)
 
 ## The WunderGraph Stack
 

cli/commands/node.go

@@ -151,6 +151,7 @@ func StartWunderGraphNode(n *node.Node, opts ...Option) error {
 		node.WithStaticWunderNodeConfig(wunderNodeConfig),
 		node.WithForceHttpsRedirects(!disableForceHttpsRedirects),
 		node.WithIntrospection(enableIntrospection),
+		node.WithTraceBatchTimeout(otelBatchTimeout),
 	}
 
 	if shutdownAfterIdle > 0 {

cli/commands/root.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"io/fs"
 	"os"
+	"strconv"
 	"time"
 
 	"github.com/fatih/color"
@@ -41,6 +42,9 @@ var (
 	cmdDurationMetric     telemetry.DurationMetric
 	zapLogLevel           zapcore.Level
 	_wunderGraphDirConfig string
+	// otelBatchTimeout is the maximum timeout before a batch of otel data is sent
+	// By default it is 5 seconds but for CI and debugging purposes it should be set much lower
+	otelBatchTimeout time.Duration
 
 	rootFlags helpers.RootFlags
 
@@ -268,6 +272,15 @@ func init() {
 	_, isTelemetryDebugEnabled := os.LookupEnv("WG_TELEMETRY_DEBUG")
 	telemetryAnonymousID := os.Getenv("WG_TELEMETRY_ANONYMOUS_ID")
 
+	if otelBatchTimeoutEnv, ok := os.LookupEnv("WG_OTEL_BATCH_TIMEOUT_MS"); ok {
+		ms, err := strconv.Atoi(otelBatchTimeoutEnv)
+		if err != nil {
+			log.Error("Value behind WG_OTEL_BATCH_TIMEOUT_MS could not be parsed as integer", zap.Error(err))
+		} else {
+			otelBatchTimeout = time.Duration(ms) * time.Millisecond
+		}
+	}
+
 	config.InitConfig(config.Options{
 		TelemetryEnabled:     !isTelemetryDisabled,
 		TelemetryAnonymousID: telemetryAnonymousID,

cli/commands/start.go

@@ -66,7 +66,7 @@ var startCmd = &cobra.Command{
 		// Only exit with error code 1 when the server was not stopped by the signal
 		if err := g.Wait(); sigCtx.Err() == nil && err != nil {
 			// Exit with error code 1 to indicate failure and restart
-			log.Error("WunderGraph start process shutdown: %w", zap.Error(err))
+			log.Error("WunderGraph start process shutdown", zap.Error(err))
 			return err
 		}