Insert continuous-test as a module (#7747)

CHANGELOG.md

@@ -7,6 +7,7 @@
 * [CHANGE] Querier: the CLI flag `-querier.minimize-ingester-requests` has been moved from "experimental" to "advanced". #7638
 * [CHANGE] Ingester: `/ingester/flush` endpoint is now only allowed to execute only while the ingester is in `Running` state. The 503 status code is returned if the endpoint is called while the ingester is not in `Running` state. #7486
 * [CHANGE] Distributor: Include label name in `err-mimir-label-value-too-long` error message: #7740
+* [FEATURE] Continuous-test: now runable as a module with `mimir -target=continuous-test`. #7747
 * [FEATURE] Store-gateway: Allow specific tenants to be enabled or disabled via `-store-gateway.enabled-tenants` or `-store-gateway.disabled-tenants` CLI flags or their corresponding YAML settings. #7653
 * [FEATURE] New `-<prefix>.s3.bucket-lookup-type` flag configures lookup style type, used to access bucket in s3 compatible providers. #7684
 * [ENHANCEMENT] Store-gateway: merge series from different blocks concurrently. #7456

cmd/mimir-continuous-test/main.go

@@ -22,33 +22,23 @@ import (
 	"github.com/grafana/mimir/pkg/util/version"
 )
 
-type Config struct {
-	ServerMetricsPort   int
-	LogLevel            log.Level
-	Client              continuoustest.ClientConfig
-	Manager             continuoustest.ManagerConfig
-	WriteReadSeriesTest continuoustest.WriteReadSeriesTestConfig
-}
-
-func (cfg *Config) RegisterFlags(f *flag.FlagSet) {
-	f.IntVar(&cfg.ServerMetricsPort, "server.metrics-port", 9900, "The port where metrics are exposed.")
-	cfg.LogLevel.RegisterFlags(f)
-	cfg.Client.RegisterFlags(f)
-	cfg.Manager.RegisterFlags(f)
-	cfg.WriteReadSeriesTest.RegisterFlags(f)
-}
-
 func main() {
 	// Parse CLI arguments.
-	cfg := &Config{}
+	cfg := &continuoustest.Config{}
+	var (
+		serverMetricsPort int
+		logLevel          log.Level
+	)
+	flag.CommandLine.IntVar(&serverMetricsPort, "server.metrics-port", 9900, "The port where metrics are exposed.")
 	cfg.RegisterFlags(flag.CommandLine)
+	logLevel.RegisterFlags(flag.CommandLine)
 
 	if err := flagext.ParseFlagsWithoutArguments(flag.CommandLine); err != nil {
 		fmt.Fprintln(os.Stderr, err.Error())
 		os.Exit(1)
 	}
 
-	util_log.InitLogger(log.LogfmtFormat, cfg.LogLevel, false, util_log.RateLimitedLoggerCfg{})
+	util_log.InitLogger(log.LogfmtFormat, logLevel, false, util_log.RateLimitedLoggerCfg{})
 
 	// Setting the environment variable JAEGER_AGENT_HOST enables tracing.
 	if trace, err := tracing.NewFromEnv("mimir-continuous-test", jaegercfg.MaxTagValueLength(16e3)); err != nil {
@@ -64,7 +54,7 @@ func main() {
 	registry.MustRegister(version.NewCollector("mimir_continuous_test"))
 	registry.MustRegister(collectors.NewGoCollector())
 
-	i := instrumentation.NewMetricsServer(cfg.ServerMetricsPort, registry)
+	i := instrumentation.NewMetricsServer(serverMetricsPort, registry)
 	if err := i.Start(); err != nil {
 		level.Error(logger).Log("msg", "Unable to start instrumentation server", "err", err.Error())
 		util_log.Flush()

cmd/mimir/help-all.txt.tmpl

@@ -2765,6 +2765,38 @@ Usage of ./cmd/mimir/mimir:
     	[experimental] The number of workers used for each tenant federated query. This setting limits the maximum number of per-tenant queries executed at a time for a tenant federated query. (default 16)
   -tenant-federation.max-tenants int
     	[experimental] The max number of tenant IDs that may be supplied for a federated query if enabled. 0 to disable the limit.
+  -tests.basic-auth-password string
+    	The password to use for HTTP bearer authentication. (mutually exclusive with tenant-id or bearer-token flags)
+  -tests.basic-auth-user string
+    	The username to use for HTTP bearer authentication. (mutually exclusive with tenant-id or bearer-token flags)
+  -tests.bearer-token string
+    	The bearer token to use for HTTP bearer authentication. (mutually exclusive with tenant-id flag or basic-auth flags)
+  -tests.read-endpoint string
+    	The base endpoint on the read path. The URL should have no trailing slash. The specific API path is appended by the tool to the URL, for example /api/v1/query_range for range query API, so the configured URL must not include it.
+  -tests.read-timeout duration
+    	The timeout for a single read request. (default 1m0s)
+  -tests.run-interval duration
+    	How frequently tests should run. (default 5m0s)
+  -tests.smoke-test
+    	Run a smoke test, i.e. run all tests once and exit.
+  -tests.tenant-id string
+    	The tenant ID to use to write and read metrics in tests. (mutually exclusive with basic-auth or bearer-token flags) (default "anonymous")
+  -tests.write-batch-size int
+    	The maximum number of series to write in a single request. (default 1000)
+  -tests.write-endpoint string
+    	The base endpoint on the write path. The URL should have no trailing slash. The specific API path is appended by the tool to the URL, for example /api/v1/push for the remote write API endpoint, so the configured URL must not include it.
+  -tests.write-protocol string
+    	The protocol to use to write series data. Supported values are: prometheus, otlp-http (default "prometheus")
+  -tests.write-read-series-test.float-samples-enabled
+    	Set to true to use float samples (default true)
+  -tests.write-read-series-test.histogram-samples-enabled
+    	Set to true to use native histogram samples
+  -tests.write-read-series-test.max-query-age duration
+    	How back in the past metrics can be queried at most. (default 168h0m0s)
+  -tests.write-read-series-test.num-series int
+    	Number of series used for the test. (default 10000)
+  -tests.write-timeout duration
+    	The timeout for a single write request. (default 5s)
   -timeseries-unmarshal-caching-optimization-enabled
     	[experimental] Enables optimized marshaling of timeseries. (default true)
   -usage-stats.enabled

cmd/mimir/help.txt.tmpl

@@ -719,6 +719,38 @@ Usage of ./cmd/mimir/mimir:
     	Comma-separated list of components to include in the instantiated process. The default value 'all' includes all components that are required to form a functional Grafana Mimir instance in single-binary mode. Use the '-modules' command line flag to get a list of available components, and to see which components are included with 'all'. (default all)
   -tenant-federation.enabled
     	If enabled on all services, queries can be federated across multiple tenants. The tenant IDs involved need to be specified separated by a '|' character in the 'X-Scope-OrgID' header.
+  -tests.basic-auth-password string
+    	The password to use for HTTP bearer authentication. (mutually exclusive with tenant-id or bearer-token flags)
+  -tests.basic-auth-user string
+    	The username to use for HTTP bearer authentication. (mutually exclusive with tenant-id or bearer-token flags)
+  -tests.bearer-token string
+    	The bearer token to use for HTTP bearer authentication. (mutually exclusive with tenant-id flag or basic-auth flags)
+  -tests.read-endpoint string
+    	The base endpoint on the read path. The URL should have no trailing slash. The specific API path is appended by the tool to the URL, for example /api/v1/query_range for range query API, so the configured URL must not include it.
+  -tests.read-timeout duration
+    	The timeout for a single read request. (default 1m0s)
+  -tests.run-interval duration
+    	How frequently tests should run. (default 5m0s)
+  -tests.smoke-test
+    	Run a smoke test, i.e. run all tests once and exit.
+  -tests.tenant-id string
+    	The tenant ID to use to write and read metrics in tests. (mutually exclusive with basic-auth or bearer-token flags) (default "anonymous")
+  -tests.write-batch-size int
+    	The maximum number of series to write in a single request. (default 1000)
+  -tests.write-endpoint string
+    	The base endpoint on the write path. The URL should have no trailing slash. The specific API path is appended by the tool to the URL, for example /api/v1/push for the remote write API endpoint, so the configured URL must not include it.
+  -tests.write-protocol string
+    	The protocol to use to write series data. Supported values are: prometheus, otlp-http (default "prometheus")
+  -tests.write-read-series-test.float-samples-enabled
+    	Set to true to use float samples (default true)
+  -tests.write-read-series-test.histogram-samples-enabled
+    	Set to true to use native histogram samples
+  -tests.write-read-series-test.max-query-age duration
+    	How back in the past metrics can be queried at most. (default 168h0m0s)
+  -tests.write-read-series-test.num-series int
+    	Number of series used for the test. (default 10000)
+  -tests.write-timeout duration
+    	The timeout for a single write request. (default 5s)
   -usage-stats.enabled
     	Enable anonymous usage reporting. (default true)
   -usage-stats.installation-mode string

pkg/continuoustest/config.go

@@ -0,0 +1,19 @@
+// SPDX-License-Identifier: AGPL-3.0-only
+
+package continuoustest
+
+import (
+	"flag"
+)
+
+type Config struct {
+	Client              ClientConfig              `yaml:"-"`
+	Manager             ManagerConfig             `yaml:"-"`
+	WriteReadSeriesTest WriteReadSeriesTestConfig `yaml:"-"`
+}
+
+func (cfg *Config) RegisterFlags(f *flag.FlagSet) {
+	cfg.Client.RegisterFlags(f)
+	cfg.Manager.RegisterFlags(f)
+	cfg.WriteReadSeriesTest.RegisterFlags(f)
+}

pkg/mimir/mimir.go

@@ -48,6 +48,7 @@ import (
 	alertstorelocal "github.com/grafana/mimir/pkg/alertmanager/alertstore/local"
 	"github.com/grafana/mimir/pkg/api"
 	"github.com/grafana/mimir/pkg/compactor"
+	"github.com/grafana/mimir/pkg/continuoustest"
 	"github.com/grafana/mimir/pkg/distributor"
 	"github.com/grafana/mimir/pkg/flusher"
 	"github.com/grafana/mimir/pkg/frontend"
@@ -135,6 +136,7 @@ type Config struct {
 	MemberlistKV        memberlist.KVConfig                        `yaml:"memberlist"`
 	QueryScheduler      scheduler.Config                           `yaml:"query_scheduler"`
 	UsageStats          usagestats.Config                          `yaml:"usage_stats"`
+	ContinuousTest      continuoustest.Config                      `yaml:"-"`
 	OverridesExporter   exporter.Config                            `yaml:"overrides_exporter"`
 
 	Common CommonConfig `yaml:"common"`
@@ -192,6 +194,7 @@ func (c *Config) RegisterFlags(f *flag.FlagSet, logger log.Logger) {
 	c.ActivityTracker.RegisterFlags(f)
 	c.QueryScheduler.RegisterFlags(f, logger)
 	c.UsageStats.RegisterFlags(f)
+	c.ContinuousTest.RegisterFlags(f)
 	c.OverridesExporter.RegisterFlags(f, logger)
 
 	c.Common.RegisterFlags(f)
@@ -720,6 +723,7 @@ type Mimir struct {
 	ActivityTracker               *activitytracker.ActivityTracker
 	Vault                         *vault.Vault
 	UsageStatsReporter            *usagestats.Reporter
+	ContinuousTestManager         *continuoustest.Manager
 	BuildInfoHandler              http.Handler
 }
 

pkg/mimir/modules.go

@@ -40,6 +40,7 @@ import (
 	"github.com/grafana/mimir/pkg/alertmanager/alertstore"
 	"github.com/grafana/mimir/pkg/api"
 	"github.com/grafana/mimir/pkg/compactor"
+	"github.com/grafana/mimir/pkg/continuoustest"
 	"github.com/grafana/mimir/pkg/distributor"
 	"github.com/grafana/mimir/pkg/flusher"
 	"github.com/grafana/mimir/pkg/frontend"
@@ -97,6 +98,7 @@ const (
 	Vault                      string = "vault"
 	TenantFederation           string = "tenant-federation"
 	UsageStats                 string = "usage-stats"
+	ContinuousTest             string = "continuous-test"
 	All                        string = "all"
 
 	// Write Read and Backend are the targets used when using the read-write deployment mode.
@@ -1034,6 +1036,20 @@ func (t *Mimir) initUsageStats() (services.Service, error) {
 	return t.UsageStatsReporter, nil
 }
 
+func (t *Mimir) initContinuousTest() (services.Service, error) {
+	client, err := continuoustest.NewClient(t.Cfg.ContinuousTest.Client, util_log.Logger)
+	if err != nil {
+		return nil, errors.Wrap(err, "Failed to initialize continuous-test client")
+	}
+
+	t.ContinuousTestManager = continuoustest.NewManager(t.Cfg.ContinuousTest.Manager, util_log.Logger)
+	t.ContinuousTestManager.AddTest(continuoustest.NewWriteReadSeriesTest(t.Cfg.ContinuousTest.WriteReadSeriesTest, client, util_log.Logger, t.Registerer))
+
+	return services.NewBasicService(nil, func(ctx context.Context) error {
+		return t.ContinuousTestManager.Run(ctx)
+	}, nil), nil
+}
+
 func (t *Mimir) setupModuleManager() error {
 	mm := modules.NewManager(util_log.Logger)
 
@@ -1069,6 +1085,7 @@ func (t *Mimir) setupModuleManager() error {
 	mm.RegisterModule(QueryScheduler, t.initQueryScheduler)
 	mm.RegisterModule(TenantFederation, t.initTenantFederation, modules.UserInvisibleModule)
 	mm.RegisterModule(UsageStats, t.initUsageStats, modules.UserInvisibleModule)
+	mm.RegisterModule(ContinuousTest, t.initContinuousTest)
 	mm.RegisterModule(Vault, t.initVault, modules.UserInvisibleModule)
 	mm.RegisterModule(Write, nil)
 	mm.RegisterModule(Read, nil)
@@ -1102,6 +1119,7 @@ func (t *Mimir) setupModuleManager() error {
 		Compactor:                {API, MemberlistKV, Overrides, Vault},
 		StoreGateway:             {API, Overrides, MemberlistKV, Vault},
 		TenantFederation:         {Queryable},
+		ContinuousTest:           {API},
 		Write:                    {Distributor, Ingester},
 		Read:                     {QueryFrontend, Querier},
 		Backend:                  {QueryScheduler, Ruler, StoreGateway, Compactor, AlertManager, OverridesExporter},