Merge pull request #380 from pace/sentry-tracing-poc

tracing: replace Jaeger with Sentry

Author: monstermunchkin

.github/workflows/golang-ci.yml

@@ -6,7 +6,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        go: [ '1.21.12' ]
+        go: [ '1.23' ]
     steps:
     - uses: actions/checkout@v2
     - uses: actions/setup-go@v2

README.md

@@ -14,7 +14,7 @@ A pace/bricks microservice is:
 * configured using **environment variables** ([TWELVE-FACTOR APP](https://12factor.net/))
 * monitored using **prometheus**
 * reports errors to **sentry**
-* samples traces to **jaeger**
+* samples traces to **sentry**
 * **logs** to stdout using json deployed **kubernetes** otherwise human readable
 * offers **health** endpoints
 * built-in redaction of JWTs and card schemes

backend/couchdb/db.go

@@ -79,7 +79,7 @@ func Client(cfg *Config) (*kivik.Client, error) {
 			Username: cfg.User,
 			Password: cfg.Password,
 		},
-		&transport.JaegerRoundTripper{},
+		&transport.TracingRoundTripper{},
 		transport.NewDumpRoundTripperEnv(),
 	}
 	if !cfg.DisableRequestLogging {

backend/postgres/postgres.go

@@ -14,8 +14,8 @@ import (
 	"sync"
 	"time"
 
-	"github.com/opentracing/opentracing-go"
-	olog "github.com/opentracing/opentracing-go/log"
+	"github.com/getsentry/sentry-go"
+
 	"github.com/rs/zerolog"
 
 	"github.com/caarlos0/env/v10"
@@ -238,7 +238,7 @@ func CustomConnectionPool(opts *pg.Options) *pg.DB {
 		log.Logger().Warn().Msg("Connection pool has logging queries disabled completely")
 	}
 
-	db.OnQueryProcessed(openTracingAdapter)
+	db.OnQueryProcessed(tracingAdapter)
 	db.OnQueryProcessed(func(event *pg.QueryProcessedEvent) {
 		metricsAdapter(event, opts)
 	})
@@ -329,38 +329,34 @@ func getQueryType(s string) string {
 	return strings.ToUpper(s)
 }
 
-func openTracingAdapter(event *pg.QueryProcessedEvent) {
+func tracingAdapter(event *pg.QueryProcessedEvent) {
 	// start span with general info
 	q, qe := event.UnformattedQuery()
 	if qe != nil {
 		// this is only a display issue not a "real" issue
 		q = qe.Error()
 	}
 
-	span, _ := opentracing.StartSpanFromContext(event.DB.Context(), "sql: "+getQueryType(q),
-		opentracing.StartTime(event.StartTime))
+	span := sentry.StartSpan(event.DB.Context(), "db.sql.query", sentry.WithDescription(getQueryType(q)))
+	defer span.Finish()
+
+	span.StartTime = event.StartTime
 
 	span.SetTag("db.system", "postgres")
 
-	fields := []olog.Field{
-		olog.String("file", event.File),
-		olog.Int("line", event.Line),
-		olog.String("func", event.Func),
-		olog.Int("attempt", event.Attempt),
-		olog.String("query", q),
-	}
+	span.SetData("file", event.File)
+	span.SetData("line", event.Line)
+	span.SetData("func", event.Func)
+	span.SetData("attempt", event.Attempt)
+	span.SetData("query", q)
 
 	// add error or result set info
 	if event.Error != nil {
-		fields = append(fields, olog.Error(event.Error))
+		span.SetData("error", event.Error)
 	} else {
-		fields = append(fields,
-			olog.Int("affected", event.Result.RowsAffected()),
-			olog.Int("rows", event.Result.RowsReturned()))
+		span.SetData("affected", event.Result.RowsAffected())
+		span.SetData("rows", event.Result.RowsReturned())
 	}
-
-	span.LogFields(fields...)
-	span.Finish()
 }
 
 func metricsAdapter(event *pg.QueryProcessedEvent, opts *pg.Options) {

backend/redis/redis.go

@@ -8,12 +8,12 @@ import (
 	"time"
 
 	"github.com/caarlos0/env/v10"
-	"github.com/opentracing/opentracing-go"
-	olog "github.com/opentracing/opentracing-go/log"
-	"github.com/pace/bricks/maintenance/health/servicehealthcheck"
-	"github.com/pace/bricks/maintenance/log"
+	"github.com/getsentry/sentry-go"
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/redis/go-redis/v9"
+
+	"github.com/pace/bricks/maintenance/health/servicehealthcheck"
+	"github.com/pace/bricks/maintenance/log"
 )
 
 type config struct {
@@ -157,7 +157,7 @@ type logtracerKey struct{}
 
 type logtracerValues struct {
 	startedAt time.Time
-	span      opentracing.Span
+	span      *sentry.Span
 }
 
 func (lt *logtracer) DialHook(next redis.DialHook) redis.DialHook {
@@ -168,12 +168,13 @@ func (lt *logtracer) ProcessHook(next redis.ProcessHook) redis.ProcessHook {
 	return func(ctx context.Context, cmd redis.Cmder) error {
 		startedAt := time.Now()
 
-		span, _ := opentracing.StartSpanFromContext(ctx, "redis: "+cmd.Name())
-		span.LogFields(olog.String("cmd", cmd.Name()))
+		span := sentry.StartSpan(ctx, "db.redis", sentry.WithDescription(cmd.Name()))
 		defer span.Finish()
 
 		span.SetTag("db.system", "redis")
 
+		span.SetData("cmd", cmd.Name())
+
 		paceRedisCmdTotal.With(prometheus.Labels{
 			"method": cmd.Name(),
 		}).Inc()
@@ -191,7 +192,7 @@ func (lt *logtracer) ProcessHook(next redis.ProcessHook) redis.ProcessHook {
 		// add error
 		cmdErr := cmd.Err()
 		if cmdErr != nil {
-			vals.span.LogFields(olog.Error(cmdErr))
+			vals.span.SetData("error", cmdErr)
 			le = le.Err(cmdErr)
 			paceRedisCmdFailed.With(prometheus.Labels{
 				"method": cmd.Name(),

docker-compose.yml

@@ -25,7 +25,7 @@ services:
       - MINIO_ACCESS_KEY=client
       - MINIO_SECRET_KEY=secret01
       - MINIO_REGION_NAME=us-east-1
-  
+
   couchdb:
     image: couchdb
     volumes:
@@ -56,7 +56,7 @@ services:
      - "./prometheus.yml:/etc/prometheus/prometheus.yml"
 
   testserver:
-    image: "golang:1.21"
+    image: "golang:1.23"
     volumes:
       - .:/srv
     working_dir: /srv