Merge pull request #14705 from prometheus/owilliams/default-on

utf8: enable utf-8 support by default

cmd/prometheus/main.go

@@ -101,6 +101,8 @@ var (
 )
 
 func init() {
+	// This can be removed when the default validation scheme in common is updated.
+	model.NameValidationScheme = model.UTF8Validation
 	prometheus.MustRegister(versioncollector.NewCollector(strings.ReplaceAll(appName, "-", "_")))
 
 	var err error
@@ -231,9 +233,6 @@ func (c *flagConfig) setFeatureListOptions(logger log.Logger) error {
 			case "promql-delayed-name-removal":
 				c.promqlEnableDelayedNameRemoval = true
 				level.Info(logger).Log("msg", "Experimental PromQL delayed name removal enabled.")
-			case "utf8-names":
-				model.NameValidationScheme = model.UTF8Validation
-				level.Info(logger).Log("msg", "Experimental UTF-8 support enabled")
 			case "":
 				continue
 			default:

cmd/prometheus/main_test.go

@@ -42,6 +42,11 @@ import (
 	"github.com/prometheus/prometheus/rules"
 )
 
+func init() {
+	// This can be removed when the default validation scheme in common is updated.
+	model.NameValidationScheme = model.UTF8Validation
+}
+
 const startupTime = 10 * time.Second
 
 var (

cmd/promtool/main.go

@@ -62,6 +62,11 @@ import (
 	"github.com/prometheus/prometheus/util/documentcli"
 )
 
+func init() {
+	// This can be removed when the default validation scheme in common is updated.
+	model.NameValidationScheme = model.UTF8Validation
+}
+
 const (
 	successExitCode = 0
 	failureExitCode = 1

cmd/promtool/main_test.go

@@ -31,12 +31,18 @@ import (
 	"testing"
 	"time"
 
+	"github.com/prometheus/common/model"
 	"github.com/stretchr/testify/require"
 
 	"github.com/prometheus/prometheus/model/labels"
 	"github.com/prometheus/prometheus/model/rulefmt"
 )
 
+func init() {
+	// This can be removed when the default validation scheme in common is updated.
+	model.NameValidationScheme = model.UTF8Validation
+}
+
 var promtoolPath = os.Args[0]
 
 func TestMain(m *testing.M) {

config/config.go

@@ -774,10 +774,10 @@ func (c *ScrapeConfig) Validate(globalConfig GlobalConfig) error {
 	}
 
 	switch globalConfig.MetricNameValidationScheme {
-	case "", LegacyValidationConfig:
+	case LegacyValidationConfig:
-	case UTF8ValidationConfig:
+	case "", UTF8ValidationConfig:
 		if model.NameValidationScheme != model.UTF8Validation {
-			return fmt.Errorf("utf8 name validation requested but feature not enabled via --enable-feature=utf8-names")
+			panic("utf8 name validation requested but model.NameValidationScheme is not set to UTF8")
 		}
 	default:
 		return fmt.Errorf("unknown name validation method specified, must be either 'legacy' or 'utf8', got %s", globalConfig.MetricNameValidationScheme)

config/config_test.go

@@ -62,6 +62,11 @@ import (
 	"github.com/prometheus/prometheus/util/testutil"
 )
 
+func init() {
+	// This can be removed when the default validation scheme in common is updated.
+	model.NameValidationScheme = model.UTF8Validation
+}
+
 func mustParseURL(u string) *config.URL {
 	parsed, err := url.Parse(u)
 	if err != nil {
@@ -2042,6 +2047,10 @@ var expectedErrors = []struct {
 }
 
 func TestBadConfigs(t *testing.T) {
+	model.NameValidationScheme = model.LegacyValidation
+	defer func() {
+		model.NameValidationScheme = model.UTF8Validation
+	}()
 	for _, ee := range expectedErrors {
 		_, err := LoadFile("testdata/"+ee.filename, false, false, log.NewNopLogger())
 		require.Error(t, err, "%s", ee.filename)
@@ -2051,6 +2060,10 @@ func TestBadConfigs(t *testing.T) {
 }
 
 func TestBadStaticConfigsJSON(t *testing.T) {
+	model.NameValidationScheme = model.LegacyValidation
+	defer func() {
+		model.NameValidationScheme = model.UTF8Validation
+	}()
 	content, err := os.ReadFile("testdata/static_config.bad.json")
 	require.NoError(t, err)
 	var tg targetgroup.Group
@@ -2059,6 +2072,10 @@ func TestBadStaticConfigsJSON(t *testing.T) {
 }
 
 func TestBadStaticConfigsYML(t *testing.T) {
+	model.NameValidationScheme = model.LegacyValidation
+	defer func() {
+		model.NameValidationScheme = model.UTF8Validation
+	}()
 	content, err := os.ReadFile("testdata/static_config.bad.yml")
 	require.NoError(t, err)
 	var tg targetgroup.Group
@@ -2323,17 +2340,17 @@ func TestScrapeConfigNameValidationSettings(t *testing.T) {
 		{
 			name:         "global setting implies local settings",
 			inputFile:    "scrape_config_global_validation_mode",
-			expectScheme: "utf8",
+			expectScheme: "legacy",
 		},
 		{
 			name:         "local setting",
 			inputFile:    "scrape_config_local_validation_mode",
-			expectScheme: "utf8",
+			expectScheme: "legacy",
 		},
 		{
 			name:         "local setting overrides global setting",
 			inputFile:    "scrape_config_local_global_validation_mode",
-			expectScheme: "legacy",
+			expectScheme: "utf8",
 		},
 	}
 

config/testdata/jobname_dup.bad.yml

@@ -1,4 +1,6 @@
 # Two scrape configs with the same job names are not allowed.
+global:
+  metric_name_validation_scheme: legacy
 scrape_configs:
   - job_name: prometheus
   - job_name: service-x

config/testdata/lowercase.bad.yml

@@ -1,3 +1,5 @@
+global:
+  metric_name_validation_scheme: legacy
 scrape_configs:
   - job_name: prometheus
     relabel_configs:

config/testdata/scrape_config_global_validation_mode.yml

@@ -1,4 +1,4 @@
 global:
-  metric_name_validation_scheme: utf8
+  metric_name_validation_scheme: legacy
 scrape_configs:
   - job_name: prometheus

config/testdata/scrape_config_local_global_validation_mode.yml

@@ -1,5 +1,5 @@
 global:
-  metric_name_validation_scheme: utf8
+  metric_name_validation_scheme: legacy
 scrape_configs:
   - job_name: prometheus
-    metric_name_validation_scheme: legacy
+    metric_name_validation_scheme: utf8

config/testdata/scrape_config_local_validation_mode.yml

@@ -1,3 +1,3 @@
 scrape_configs:
   - job_name: prometheus
-    metric_name_validation_scheme: utf8
+    metric_name_validation_scheme: legacy

docs/configuration/configuration.md

@@ -122,9 +122,9 @@ global:
   [ keep_dropped_targets: <int> | default = 0 ]
 
   # Specifies the validation scheme for metric and label names. Either blank or
-  # "legacy" for letters, numbers, colons, and underscores; or "utf8" for full
+  # "utf8" for for full UTF-8 support, or "legacy" for letters, numbers, colons,
-  # UTF-8 support.
+  # and underscores.
-  [ metric_name_validation_scheme <string> | default "legacy" ]
+  [ metric_name_validation_scheme <string> | default "utf8" ]
 
 runtime:
   # Configure the Go garbage collector GOGC parameter
@@ -477,10 +477,10 @@ metric_relabel_configs:
 # that will be kept in memory. 0 means no limit.
 [ keep_dropped_targets: <int> | default = 0 ]
 
-# Specifies the validation scheme for metric and label names. Either blank or
+# Specifies the validation scheme for metric and label names. Either blank or 
-# "legacy" for letters, numbers, colons, and underscores; or "utf8" for full
+# "utf8" for full UTF-8 support, or "legacy" for letters, numbers, colons, and
-# UTF-8 support.
+# underscores.
-[ metric_name_validation_scheme <string> | default "legacy" ]
+[ metric_name_validation_scheme <string> | default "utf8" ]
 
 # Limit on total number of positive and negative buckets allowed in a single
 # native histogram. The resolution of a histogram with more buckets will be

docs/feature_flags.md

@@ -236,10 +236,3 @@ When enabled, Prometheus will change the way in which the `__name__` label is re
 
 This allows optionally preserving the `__name__` label via the `label_replace` and `label_join` functions, and helps prevent the "vector cannot contain metrics with the same labelset" error, which can happen when applying a regex-matcher to the `__name__` label.
 
-## UTF-8 Name Support
-
-`--enable-feature=utf8-names`
-
-When enabled, changes the metric and label name validation scheme inside Prometheus to allow the full UTF-8 character set.
-By itself, this flag does not enable the request of UTF-8 names via content negotiation.
-Users will also have to set `metric_name_validation_scheme` in scrape configs to enable the feature either on the global config or on a per-scrape config basis.

scrape/manager_test.go

@@ -45,6 +45,11 @@ import (
 	"github.com/prometheus/prometheus/util/testutil"
 )
 
+func init() {
+	// This can be removed when the default validation scheme in common is updated.
+	model.NameValidationScheme = model.UTF8Validation
+}
+
 func TestPopulateLabels(t *testing.T) {
 	cases := []struct {
 		in            labels.Labels

scrape/scrape.go

@@ -305,9 +305,9 @@ func (sp *scrapePool) restartLoops(reuseCache bool) {
 		mrc                      = sp.config.MetricRelabelConfigs
 	)
 
-	validationScheme := model.LegacyValidation
+	validationScheme := model.UTF8Validation
-	if sp.config.MetricNameValidationScheme == config.UTF8ValidationConfig {
+	if sp.config.MetricNameValidationScheme == config.LegacyValidationConfig {
-		validationScheme = model.UTF8Validation
+		validationScheme = model.LegacyValidation
 	}
 
 	sp.targetMtx.Lock()
@@ -460,9 +460,9 @@ func (sp *scrapePool) sync(targets []*Target) {
 		scrapeClassicHistograms  = sp.config.ScrapeClassicHistograms
 	)
 
-	validationScheme := model.LegacyValidation
+	validationScheme := model.UTF8Validation
-	if sp.config.MetricNameValidationScheme == config.UTF8ValidationConfig {
+	if sp.config.MetricNameValidationScheme == config.LegacyValidationConfig {
-		validationScheme = model.UTF8Validation
+		validationScheme = model.LegacyValidation
 	}
 
 	sp.targetMtx.Lock()

scrape/scrape_test.go

@@ -1040,6 +1040,7 @@ func TestScrapeLoopSeriesAdded(t *testing.T) {
 }
 
 func TestScrapeLoopFailWithInvalidLabelsAfterRelabel(t *testing.T) {
+	model.NameValidationScheme = model.LegacyValidation
 	s := teststorage.New(t)
 	defer s.Close()
 	ctx, cancel := context.WithCancel(context.Background())
@@ -3768,6 +3769,7 @@ func testNativeHistogramMaxSchemaSet(t *testing.T, minBucketFactor string, expec
 	// Create a scrape loop with the HTTP server as the target.
 	configStr := fmt.Sprintf(`
 global:
+  metric_name_validation_scheme: legacy
   scrape_interval: 1s
   scrape_timeout: 1s
 scrape_configs: