Add influx push endpoint to mimir

CHANGELOG.md

@@ -2,6 +2,8 @@
 
 ## main / unreleased
 
+* [FEATURE] Distributor: Add experimental Influx handler. #10153
+
 ### Grafana Mimir
 
 * [CHANGE] Distributor: OTLP and push handler replace all non-UTF8 characters with the unicode replacement character `\uFFFD` in error messages before propagating them. #10236

docs/sources/mimir/configure/about-versioning.md

@@ -74,6 +74,10 @@ The following features are currently experimental:
   - Cache rule group contents.
     - `-ruler-storage.cache.rule-group-enabled`
 - Distributor
+  - Influx ingestion
+    - `/api/v1/influx/push` endpoint
+    - `-distributor.influx-endpoint-enabled`
+    - `-distributor.max-influx-request-size`
   - Metrics relabeling
     - `-distributor.metric-relabeling-enabled`
   - Using status code 529 instead of 429 upon rate limit exhaustion.

go.mod

@@ -25,6 +25,7 @@ require (
 	github.com/grafana/dskit v0.0.0-20250107142522-441a90acd4e5
 	github.com/grafana/e2e v0.1.2-0.20240118170847-db90b84177fc
 	github.com/hashicorp/golang-lru v1.0.2 // indirect
+	github.com/influxdata/influxdb/v2 v2.7.11
 	github.com/json-iterator/go v1.1.12
 	github.com/minio/minio-go/v7 v7.0.83
 	github.com/mitchellh/go-wordwrap v1.0.1

go.sum

@@ -1394,6 +1394,8 @@ github.com/imdario/mergo v0.3.16 h1:wwQJbIsHYGMUyLSPrEq1CT16AhnhNJQ51+4fdHUnCl4=
 github.com/imdario/mergo v0.3.16/go.mod h1:WBLT9ZmE3lPoWsEzCh9LPo3TiwVN+ZKEjmz+hD27ysY=
 github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2s0bqwp9tc8=
 github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw=
+github.com/influxdata/influxdb/v2 v2.7.11 h1:qs9qr5hsuFrlTiBtr5lBrALbQ2dHAanf21fBLlLpKww=
+github.com/influxdata/influxdb/v2 v2.7.11/go.mod h1:zNOyzQy6WbfvGi1CK1aJ2W8khOq9+Gdsj8yLj8bHHqg=
 github.com/ionos-cloud/sdk-go/v6 v6.3.0 h1:/lTieTH9Mo/CWm3cTlFLnK10jgxjUGkAqRffGqvPteY=
 github.com/ionos-cloud/sdk-go/v6 v6.3.0/go.mod h1:SXrO9OGyWjd2rZhAhEpdYN6VUAODzzqRdqA9BCviQtI=
 github.com/jessevdk/go-flags v1.5.0 h1:1jKYvbxEjfUl0fmqTCOfonvskHHXMjBySTLW4y9LFvc=
@@ -1553,8 +1555,8 @@ github.com/onsi/gomega v1.24.0 h1:+0glovB9Jd6z3VR+ScSwQqXVTIfJcGA9UBM8yzQxhqg=
 github.com/onsi/gomega v1.24.0/go.mod h1:Z/NWtiqwBrwUt4/2loMmHL63EDLnYHmVbuBpDr2vQAg=
 github.com/opencontainers/go-digest v1.0.0 h1:apOUWs51W5PlhuyGyz9FCeeBIOUDA/6nW8Oi/yOhh5U=
 github.com/opencontainers/go-digest v1.0.0/go.mod h1:0JzlMkj0TRzQZfJkVvzbP0HBR3IKzErnv2BNG4W4MAM=
-github.com/opencontainers/image-spec v1.0.2 h1:9yCKha/T5XdGtO0q9Q9a6T5NUCsTn/DrBg0D7ufOcFM=
-github.com/opencontainers/image-spec v1.0.2/go.mod h1:BtxoFyWECRxE4U/7sNtV5W15zMzWCbyJoFRP3s7yZA0=
+github.com/opencontainers/image-spec v1.1.0-rc2 h1:2zx/Stx4Wc5pIPDvIxHXvXtQFW/7XWJGmnM7r3wg034=
+github.com/opencontainers/image-spec v1.1.0-rc2/go.mod h1:3OVijpioIKYWTqjiG0zfF6wvoJ4fAXGbjdZuI2NgsRQ=
 github.com/opentracing/opentracing-go v1.1.0/go.mod h1:UkNAQd3GIcIGf0SeVgPpRdFStlNbqXla1AfSYxPUl2o=
 github.com/opentracing/opentracing-go v1.2.1-0.20220228012449-10b1cf09e00b h1:FfH+VrHHk6Lxt9HdVS0PXzSXFyS2NbZKXv33FYPol0A=
 github.com/opentracing/opentracing-go v1.2.1-0.20220228012449-10b1cf09e00b/go.mod h1:AC62GU6hc0BrNm+9RK9VSiwa/EUe1bkIeFORAMcHvJU=

integration/e2emimir/client.go

@@ -7,6 +7,7 @@ package e2emimir
 
 import (
 	"bytes"
+	"compress/gzip"
 	"context"
 	"encoding/json"
 	"errors"
@@ -36,6 +37,7 @@ import (
 	yaml "gopkg.in/yaml.v3"
 
 	"github.com/grafana/mimir/pkg/alertmanager"
+	mimirapi "github.com/grafana/mimir/pkg/api"
 	"github.com/grafana/mimir/pkg/cardinality"
 	"github.com/grafana/mimir/pkg/distributor"
 	"github.com/grafana/mimir/pkg/frontend/querymiddleware"
@@ -187,6 +189,43 @@ func (c *Client) Push(timeseries []prompb.TimeSeries) (*http.Response, error) {
 	return res, nil
 }
 
+// PushInflux the input timeseries to the remote endpoint in Influx format.
+func (c *Client) PushInflux(timeseries []prompb.TimeSeries) (*http.Response, error) {
+	// Create write request.
+	data := distributor.TimeseriesToInfluxRequest(timeseries)
+
+	// Compress it.
+	var buf bytes.Buffer
+
+	gzipData := gzip.NewWriter(&buf)
+	if _, err := gzipData.Write([]byte(data)); err != nil {
+		return nil, err
+	}
+	if err := gzipData.Close(); err != nil {
+		return nil, err
+	}
+
+	// Create HTTP request
+	req, err := http.NewRequest("POST", fmt.Sprintf("http://%s%s", c.distributorAddress, mimirapi.InfluxPushEndpoint), &buf)
+	if err != nil {
+		return nil, err
+	}
+	req.Header.Set("Content-Encoding", "gzip")
+	req.Header.Set("X-Scope-OrgID", c.orgID)
+
+	ctx, cancel := context.WithTimeout(context.Background(), c.timeout)
+	defer cancel()
+
+	// Execute HTTP request
+	res, err := c.httpClient.Do(req.WithContext(ctx))
+	if err != nil {
+		return nil, err
+	}
+
+	defer res.Body.Close()
+	return res, nil
+}
+
 func (c *Client) PushRW2(writeRequest *promRW2.Request) (*http.Response, error) {
 	// Create write request
 	data, err := proto.Marshal(writeRequest)

integration/influx_ingestion_test.go

@@ -0,0 +1,100 @@
+// SPDX-License-Identifier: AGPL-3.0-only
+//go:build requires_docker
+
+package integration
+
+import (
+	"math"
+	"testing"
+	"time"
+
+	"github.com/grafana/e2e"
+	e2edb "github.com/grafana/e2e/db"
+	"github.com/prometheus/common/model"
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/prompb"
+	v1 "github.com/prometheus/prometheus/web/api/v1"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/grafana/mimir/integration/e2emimir"
+)
+
+func TestInfluxIngestion(t *testing.T) {
+	t.Helper()
+
+	s, err := e2e.NewScenario(networkName)
+	require.NoError(t, err)
+	defer s.Close()
+
+	// Start dependencies.
+	minio := e2edb.NewMinio(9000, blocksBucketName)
+	require.NoError(t, s.StartAndWaitReady(minio))
+
+	// Start Mimir components.
+	require.NoError(t, copyFileToSharedDir(s, "docs/configurations/single-process-config-blocks.yaml", mimirConfigFile))
+
+	// Start Mimir in single binary mode, reading the config from file and overwriting
+	// the backend config to make it work with Minio.
+	flags := mergeFlags(
+		DefaultSingleBinaryFlags(),
+		BlocksStorageFlags(),
+		BlocksStorageS3Flags(),
+		map[string]string{
+			"-distributor.influx-endpoint-enabled": "true",
+		},
+	)
+
+	mimir := e2emimir.NewSingleBinary("mimir-1", flags, e2emimir.WithConfigFile(mimirConfigFile), e2emimir.WithPorts(9009, 9095))
+	require.NoError(t, s.StartAndWaitReady(mimir))
+
+	c, err := e2emimir.NewClient(mimir.HTTPEndpoint(), mimir.HTTPEndpoint(), "", "", "user-1")
+	require.NoError(t, err)
+
+	// Push some series to Mimir.
+	now := time.Now()
+
+	series, expectedVector, expectedMatrix := generateFloatSeries("series_f1", now, prompb.Label{Name: "foo", Value: "bar"})
+	// Fix up the expectation as Influx values seem to be rounded to millionths
+	for _, s := range expectedVector {
+		s.Metric[model.LabelName("__mimir_source__")] = model.LabelValue("influx")
+		s.Value = model.SampleValue(math.Round(float64(s.Value)*1000000) / 1000000.0)
+	}
+	// Fix up the expectation as Influx values seem to be rounded to millionths
+	for im, s := range expectedMatrix {
+		for iv, v := range s.Values {
+			expectedMatrix[im].Values[iv].Value = model.SampleValue(math.Round(float64(v.Value)*1000000) / 1000000.0)
+		}
+	}
+
+	res, err := c.PushInflux(series)
+	require.NoError(t, err)
+	require.Equal(t, 204, res.StatusCode)
+
+	// Check metric to track Influx requests
+	require.NoError(t, mimir.WaitSumMetricsWithOptions(e2e.Equals(1), []string{"cortex_distributor_influx_requests_total"}, e2e.WithLabelMatchers(
+		labels.MustNewMatcher(labels.MatchEqual, "user", "user-1"))))
+
+	// Query the series.
+	result, err := c.Query("series_f1", now)
+	require.NoError(t, err)
+	require.Equal(t, model.ValVector, result.Type())
+	assert.Equal(t, expectedVector, result.(model.Vector))
+
+	labelValues, err := c.LabelValues("foo", v1.MinTime, v1.MaxTime, nil)
+	require.NoError(t, err)
+	require.Equal(t, model.LabelValues{"bar"}, labelValues)
+
+	labelNames, err := c.LabelNames(v1.MinTime, v1.MaxTime, nil)
+	require.NoError(t, err)
+	require.Equal(t, []string{"__mimir_source__", "__name__", "foo"}, labelNames)
+
+	rangeResult, err := c.QueryRange("series_f1", now.Add(-15*time.Minute), now, 15*time.Second)
+	require.NoError(t, err)
+	require.Equal(t, model.ValMatrix, rangeResult.Type())
+	require.Equal(t, expectedMatrix, rangeResult.(model.Matrix))
+
+	// No metadata to query, but we do the query anyway.
+	_, err = c.GetPrometheusMetadata()
+	require.NoError(t, err)
+}

pkg/api/api.go

@@ -256,6 +256,7 @@ func (a *API) RegisterRuntimeConfig(runtimeConfigHandler http.HandlerFunc, userL
 
 const PrometheusPushEndpoint = "/api/v1/push"
 const OTLPPushEndpoint = "/otlp/v1/metrics"
+const InfluxPushEndpoint = "/api/v1/influx/push"
 
 // RegisterDistributor registers the endpoints associated with the distributor.
 func (a *API) RegisterDistributor(d *distributor.Distributor, pushConfig distributor.Config, reg prometheus.Registerer, limits *validation.Overrides) {
@@ -265,6 +266,14 @@ func (a *API) RegisterDistributor(d *distributor.Distributor, pushConfig distrib
 		pushConfig.MaxRecvMsgSize, d.RequestBufferPool, a.sourceIPs, a.cfg.SkipLabelNameValidationHeader,
 		a.cfg.SkipLabelCountValidationHeader, limits, pushConfig.RetryConfig, d.PushWithMiddlewares, d.PushMetrics, a.logger,
 	), true, false, "POST")
+
+	if pushConfig.EnableInfluxEndpoint {
+		// The Influx Push endpoint is experimental.
+		a.RegisterRoute(InfluxPushEndpoint, distributor.InfluxHandler(
+			pushConfig.MaxInfluxRequestSize, d.RequestBufferPool, a.sourceIPs, pushConfig.RetryConfig, d.PushWithMiddlewares, d.PushMetrics, a.logger,
+		), true, false, "POST")
+	}
+
 	a.RegisterRoute(OTLPPushEndpoint, distributor.OTLPHandler(
 		pushConfig.MaxOTLPRequestSize, d.RequestBufferPool, a.sourceIPs, limits, pushConfig.OTelResourceAttributePromotionConfig,
 		pushConfig.RetryConfig, pushConfig.EnableStartTimeQuietZero, d.PushWithMiddlewares, d.PushMetrics, reg, a.logger,

pkg/distributor/distributor.go

@@ -86,7 +86,8 @@ const (
 	// metaLabelTenantID is the name of the metric_relabel_configs label with tenant ID.
 	metaLabelTenantID = model.MetaLabelPrefix + "tenant_id"
 
-	maxOTLPRequestSizeFlag = "distributor.max-otlp-request-size"
+	maxOTLPRequestSizeFlag   = "distributor.max-otlp-request-size"
+	maxInfluxRequestSizeFlag = "distributor.max-influx-request-size"
 
 	instanceIngestionRateTickInterval = time.Second
 
@@ -207,6 +208,7 @@ type Config struct {
 
 	MaxRecvMsgSize           int           `yaml:"max_recv_msg_size" category:"advanced"`
 	MaxOTLPRequestSize       int           `yaml:"max_otlp_request_size" category:"experimental"`
+	MaxInfluxRequestSize     int           `yaml:"max_influx_request_size" category:"experimental" doc:"hidden"`
 	MaxRequestPoolBufferSize int           `yaml:"max_request_pool_buffer_size" category:"experimental"`
 	RemoteTimeout            time.Duration `yaml:"remote_timeout" category:"advanced"`
 
@@ -249,6 +251,9 @@ type Config struct {
 	// OTelResourceAttributePromotionConfig allows for specializing OTel resource attribute promotion.
 	OTelResourceAttributePromotionConfig OTelResourceAttributePromotionConfig `yaml:"-"`
 
+	// Influx endpoint disabled by default
+	EnableInfluxEndpoint bool `yaml:"influx_endpoint_enabled" category:"experimental" doc:"hidden"`
+
 	// Change the implementation of OTel startTime from a real zero to a special NaN value.
 	EnableStartTimeQuietZero bool `yaml:"start_time_quiet_zero" category:"advanced" doc:"hidden"`
 }
@@ -265,9 +270,11 @@ func (cfg *Config) RegisterFlags(f *flag.FlagSet, logger log.Logger) {
 
 	f.IntVar(&cfg.MaxRecvMsgSize, "distributor.max-recv-msg-size", 100<<20, "Max message size in bytes that the distributors will accept for incoming push requests to the remote write API. If exceeded, the request will be rejected.")
 	f.IntVar(&cfg.MaxOTLPRequestSize, maxOTLPRequestSizeFlag, 100<<20, "Maximum OTLP request size in bytes that the distributors accept. Requests exceeding this limit are rejected.")
+	f.IntVar(&cfg.MaxInfluxRequestSize, maxInfluxRequestSizeFlag, 100<<20, "Maximum Influx request size in bytes that the distributors accept. Requests exceeding this limit are rejected.")
 	f.IntVar(&cfg.MaxRequestPoolBufferSize, "distributor.max-request-pool-buffer-size", 0, "Max size of the pooled buffers used for marshaling write requests. If 0, no max size is enforced.")
 	f.DurationVar(&cfg.RemoteTimeout, "distributor.remote-timeout", 2*time.Second, "Timeout for downstream ingesters.")
 	f.BoolVar(&cfg.WriteRequestsBufferPoolingEnabled, "distributor.write-requests-buffer-pooling-enabled", true, "Enable pooling of buffers used for marshaling write requests.")
+	f.BoolVar(&cfg.EnableInfluxEndpoint, "distributor.influx-endpoint-enabled", false, "Enable Influx endpoint.")
 	f.IntVar(&cfg.ReusableIngesterPushWorkers, "distributor.reusable-ingester-push-workers", 2000, "Number of pre-allocated workers used to forward push requests to the ingesters. If 0, no workers will be used and a new goroutine will be spawned for each ingester push request. If not enough workers available, new goroutine will be spawned. (Note: this is a performance optimization, not a limiting feature.)")
 	f.BoolVar(&cfg.EnableStartTimeQuietZero, "distributor.otel-start-time-quiet-zero", false, "Change the implementation of OTel startTime from a real zero to a special NaN value.")
 
@@ -293,12 +300,27 @@ const (
 )
 
 type PushMetrics struct {
+	// Influx metrics.
+	influxRequestCounter       *prometheus.CounterVec
+	influxUncompressedBodySize *prometheus.HistogramVec
+	// OTLP metrics.
 	otlpRequestCounter   *prometheus.CounterVec
 	uncompressedBodySize *prometheus.HistogramVec
 }
 
 func newPushMetrics(reg prometheus.Registerer) *PushMetrics {
 	return &PushMetrics{
+		influxRequestCounter: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
+			Name: "cortex_distributor_influx_requests_total",
+			Help: "The total number of Influx requests that have come in to the distributor.",
+		}, []string{"user"}),
+		influxUncompressedBodySize: promauto.With(reg).NewHistogramVec(prometheus.HistogramOpts{
+			Name:                            "cortex_distributor_influx_uncompressed_request_body_size_bytes",
+			Help:                            "Size of uncompressed request body in bytes.",
+			NativeHistogramBucketFactor:     1.1,
+			NativeHistogramMinResetDuration: 1 * time.Hour,
+			NativeHistogramMaxBucketNumber:  100,
+		}, []string{"user"}),
 		otlpRequestCounter: promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
 			Name: "cortex_distributor_otlp_requests_total",
 			Help: "The total number of OTLP requests that have come in to the distributor.",
@@ -313,6 +335,18 @@ func newPushMetrics(reg prometheus.Registerer) *PushMetrics {
 	}
 }
 
+func (m *PushMetrics) IncInfluxRequest(user string) {
+	if m != nil {
+		m.influxRequestCounter.WithLabelValues(user).Inc()
+	}
+}
+
+func (m *PushMetrics) ObserveInfluxUncompressedBodySize(user string, size float64) {
+	if m != nil {
+		m.influxUncompressedBodySize.WithLabelValues(user).Observe(size)
+	}
+}
+
 func (m *PushMetrics) IncOTLPRequest(user string) {
 	if m != nil {
 		m.otlpRequestCounter.WithLabelValues(user).Inc()
@@ -326,6 +360,8 @@ func (m *PushMetrics) ObserveUncompressedBodySize(user string, size float64) {
 }
 
 func (m *PushMetrics) deleteUserMetrics(user string) {
+	m.influxRequestCounter.DeleteLabelValues(user)
+	m.influxUncompressedBodySize.DeleteLabelValues(user)
 	m.otlpRequestCounter.DeleteLabelValues(user)
 	m.uncompressedBodySize.DeleteLabelValues(user)
 }