feat(kafka): Enable querier to optionally query partition ingesters (#…

…14418)

.golangci.yml

@@ -24,7 +24,7 @@ run:
     - cgo
     - promtail_journal_enabled
     - integration
-    
+
 # output configuration options
 output:
   formats:

docs/sources/shared/configuration.md

@@ -4228,6 +4228,11 @@ engine:
 # When true, querier limits sent via a header are enforced.
 # CLI flag: -querier.per-request-limits-enabled
 [per_request_limits_enabled: <boolean> | default = false]
+
+# When true, querier directs ingester queries to the partition-ingesters instead
+# of the normal ingesters.
+# CLI flag: -querier.query_partition_ingesters
+[query_partition_ingesters: <boolean> | default = false]
 ```
 
 ### query_range

go.mod

@@ -51,7 +51,7 @@ require (
 	github.com/gorilla/mux v1.8.1
 	github.com/gorilla/websocket v1.5.3
 	github.com/grafana/cloudflare-go v0.0.0-20230110200409-c627cf6792f2
-	github.com/grafana/dskit v0.0.0-20241004175247-687ec485facf
+	github.com/grafana/dskit v0.0.0-20241007172036-53283a0f6b41
 	github.com/grafana/go-gelf/v2 v2.0.1
 	github.com/grafana/gomemcache v0.0.0-20240229205252-cd6a66d6fb56
 	github.com/grafana/regexp v0.0.0-20240518133315-a468a5bfb3bc

go.sum

@@ -1044,8 +1044,8 @@ github.com/gorilla/websocket v1.5.3 h1:saDtZ6Pbx/0u+bgYQ3q96pZgCzfhKXGPqt7kZ72aN
 github.com/gorilla/websocket v1.5.3/go.mod h1:YR8l580nyteQvAITg2hZ9XVh4b55+EU/adAjf1fMHhE=
 github.com/grafana/cloudflare-go v0.0.0-20230110200409-c627cf6792f2 h1:qhugDMdQ4Vp68H0tp/0iN17DM2ehRo1rLEdOFe/gB8I=
 github.com/grafana/cloudflare-go v0.0.0-20230110200409-c627cf6792f2/go.mod h1:w/aiO1POVIeXUQyl0VQSZjl5OAGDTL5aX+4v0RA1tcw=
-github.com/grafana/dskit v0.0.0-20241004175247-687ec485facf h1:ZafqZwIpdCCMifH9Ok6C98rYaCh5OZeyyHLbU0FPedg=
-github.com/grafana/dskit v0.0.0-20241004175247-687ec485facf/go.mod h1:SPLNCARd4xdjCkue0O6hvuoveuS1dGJjDnfxYe405YQ=
+github.com/grafana/dskit v0.0.0-20241007172036-53283a0f6b41 h1:a4O59OU3FJZ+EJUVnlvvNTvdAc4uRN1P6EaGwqL9CnA=
+github.com/grafana/dskit v0.0.0-20241007172036-53283a0f6b41/go.mod h1:SPLNCARd4xdjCkue0O6hvuoveuS1dGJjDnfxYe405YQ=
 github.com/grafana/go-gelf/v2 v2.0.1 h1:BOChP0h/jLeD+7F9mL7tq10xVkDG15he3T1zHuQaWak=
 github.com/grafana/go-gelf/v2 v2.0.1/go.mod h1:lexHie0xzYGwCgiRGcvZ723bSNyNI8ZRD4s0CLobh90=
 github.com/grafana/gocql v0.0.0-20200605141915-ba5dc39ece85 h1:xLuzPoOzdfNb/RF/IENCw+oLVdZB4G21VPhkHBgwSHY=

pkg/loki/loki.go

@@ -736,7 +736,7 @@ func (t *Loki) setupModuleManager() error {
 		PatternRingClient:        {Server, MemberlistKV, Analytics},
 		PatternIngesterTee:       {Server, MemberlistKV, Analytics, PatternRingClient},
 		PatternIngester:          {Server, MemberlistKV, Analytics, PatternRingClient, PatternIngesterTee},
-		IngesterQuerier:          {Ring},
+		IngesterQuerier:          {Ring, PartitionRing, Overrides},
 		QuerySchedulerRing:       {Overrides, MemberlistKV},
 		IndexGatewayRing:         {Overrides, MemberlistKV},
 		PartitionRing:            {MemberlistKV, Server, Ring},

pkg/loki/modules.go

@@ -975,8 +975,9 @@ func (t *Loki) setupAsyncStore() error {
 }
 
 func (t *Loki) initIngesterQuerier() (_ services.Service, err error) {
-	logger := log.With(util_log.Logger, "component", "querier")
-	t.ingesterQuerier, err = querier.NewIngesterQuerier(t.Cfg.IngesterClient, t.ring, t.Cfg.Querier.ExtraQueryDelay, t.Cfg.MetricsNamespace, logger)
+	logger := log.With(util_log.Logger, "component", "ingester-querier")
+
+	t.ingesterQuerier, err = querier.NewIngesterQuerier(t.Cfg.Querier, t.Cfg.IngesterClient, t.ring, t.partitionRing, t.Overrides.IngestionPartitionsTenantShardSize, t.Cfg.MetricsNamespace, logger)
 	if err != nil {
 		return nil, err
 	}
@@ -1756,7 +1757,7 @@ func (t *Loki) initAnalytics() (services.Service, error) {
 
 // The Ingest Partition Ring is responsible for watching the available ingesters and assigning partitions to incoming requests.
 func (t *Loki) initPartitionRing() (services.Service, error) {
-	if !t.Cfg.Ingester.KafkaIngestion.Enabled {
+	if !t.Cfg.Ingester.KafkaIngestion.Enabled && !t.Cfg.Querier.QueryPartitionIngesters {
 		return nil, nil
 	}
 

pkg/querier/ingester_querier.go

@@ -8,6 +8,8 @@ import (
 
 	"github.com/go-kit/log"
 	"github.com/go-kit/log/level"
+	"github.com/grafana/dskit/concurrency"
+	"github.com/grafana/dskit/user"
 	"golang.org/x/exp/slices"
 
 	"github.com/grafana/loki/v3/pkg/storage/stores/index/seriesvolume"
@@ -40,28 +42,32 @@ type responseFromIngesters struct {
 
 // IngesterQuerier helps with querying the ingesters.
 type IngesterQuerier struct {
-	ring            ring.ReadRing
-	pool            *ring_client.Pool
-	extraQueryDelay time.Duration
-	logger          log.Logger
+	querierConfig          Config
+	ring                   ring.ReadRing
+	partitionRing          *ring.PartitionInstanceRing
+	getShardCountForTenant func(string) int
+	pool                   *ring_client.Pool
+	logger                 log.Logger
 }
 
-func NewIngesterQuerier(clientCfg client.Config, ring ring.ReadRing, extraQueryDelay time.Duration, metricsNamespace string, logger log.Logger) (*IngesterQuerier, error) {
+func NewIngesterQuerier(querierConfig Config, clientCfg client.Config, ring ring.ReadRing, partitionRing *ring.PartitionInstanceRing, getShardCountForTenant func(string) int, metricsNamespace string, logger log.Logger) (*IngesterQuerier, error) {
 	factory := func(addr string) (ring_client.PoolClient, error) {
 		return client.New(clientCfg, addr)
 	}
 
-	return newIngesterQuerier(clientCfg, ring, extraQueryDelay, ring_client.PoolAddrFunc(factory), metricsNamespace, logger)
+	return newIngesterQuerier(querierConfig, clientCfg, ring, partitionRing, getShardCountForTenant, ring_client.PoolAddrFunc(factory), metricsNamespace, logger)
 }
 
 // newIngesterQuerier creates a new IngesterQuerier and allows to pass a custom ingester client factory
 // used for testing purposes
-func newIngesterQuerier(clientCfg client.Config, ring ring.ReadRing, extraQueryDelay time.Duration, clientFactory ring_client.PoolFactory, metricsNamespace string, logger log.Logger) (*IngesterQuerier, error) {
+func newIngesterQuerier(querierConfig Config, clientCfg client.Config, ring ring.ReadRing, partitionRing *ring.PartitionInstanceRing, getShardCountForTenant func(string) int, clientFactory ring_client.PoolFactory, metricsNamespace string, logger log.Logger) (*IngesterQuerier, error) {
 	iq := IngesterQuerier{
-		ring:            ring,
-		pool:            clientpool.NewPool("ingester", clientCfg.PoolConfig, ring, clientFactory, util_log.Logger, metricsNamespace),
-		extraQueryDelay: extraQueryDelay,
-		logger:          logger,
+		querierConfig:          querierConfig,
+		ring:                   ring,
+		partitionRing:          partitionRing,
+		getShardCountForTenant: getShardCountForTenant, // limits?
+		pool:                   clientpool.NewPool("ingester", clientCfg.PoolConfig, ring, clientFactory, util_log.Logger, metricsNamespace),
+		logger:                 logger,
 	}
 
 	err := services.StartAndAwaitRunning(context.Background(), iq.pool)
@@ -73,22 +79,53 @@ func newIngesterQuerier(clientCfg client.Config, ring ring.ReadRing, extraQueryD
 }
 
 // forAllIngesters runs f, in parallel, for all ingesters
-// TODO taken from Cortex, see if we can refactor out an usable interface.
 func (q *IngesterQuerier) forAllIngesters(ctx context.Context, f func(context.Context, logproto.QuerierClient) (interface{}, error)) ([]responseFromIngesters, error) {
+	if q.querierConfig.QueryPartitionIngesters {
+		tenantID, err := user.ExtractOrgID(ctx)
+		if err != nil {
+			return nil, err
+		}
+		tenantShards := q.getShardCountForTenant(tenantID)
+		subring, err := q.partitionRing.ShuffleShardWithLookback(tenantID, tenantShards, q.querierConfig.QueryIngestersWithin, time.Now())
+		if err != nil {
+			return nil, err
+		}
+		replicationSets, err := subring.GetReplicationSetsForOperation(ring.Read)
+		if err != nil {
+			return nil, err
+		}
+		return q.forGivenIngesterSets(ctx, replicationSets, f)
+	}
+
 	replicationSet, err := q.ring.GetReplicationSetForOperation(ring.Read)
 	if err != nil {
 		return nil, err
 	}
 
-	return q.forGivenIngesters(ctx, replicationSet, f)
+	return q.forGivenIngesters(ctx, replicationSet, defaultQuorumConfig(), f)
 }
 
-// forGivenIngesters runs f, in parallel, for given ingesters
-func (q *IngesterQuerier) forGivenIngesters(ctx context.Context, replicationSet ring.ReplicationSet, f func(context.Context, logproto.QuerierClient) (interface{}, error)) ([]responseFromIngesters, error) {
-	cfg := ring.DoUntilQuorumConfig{
+// forGivenIngesterSets runs f, in parallel, for given ingester sets
+func (q *IngesterQuerier) forGivenIngesterSets(ctx context.Context, replicationSet []ring.ReplicationSet, f func(context.Context, logproto.QuerierClient) (interface{}, error)) ([]responseFromIngesters, error) {
+	// Enable minimize requests so we initially query a single ingester per replication set, as each replication-set is one partition.
+	// Ingesters must supply zone information for this to have an effect.
+	config := ring.DoUntilQuorumConfig{
+		MinimizeRequests: true,
+	}
+	return concurrency.ForEachJobMergeResults[ring.ReplicationSet, responseFromIngesters](ctx, replicationSet, 0, func(ctx context.Context, set ring.ReplicationSet) ([]responseFromIngesters, error) {
+		return q.forGivenIngesters(ctx, set, config, f)
+	})
+}
+
+func defaultQuorumConfig() ring.DoUntilQuorumConfig {
+	return ring.DoUntilQuorumConfig{
 		// Nothing here
 	}
-	results, err := ring.DoUntilQuorum(ctx, replicationSet, cfg, func(ctx context.Context, ingester *ring.InstanceDesc) (responseFromIngesters, error) {
+}
+
+// forGivenIngesters runs f, in parallel, for given ingesters
+func (q *IngesterQuerier) forGivenIngesters(ctx context.Context, replicationSet ring.ReplicationSet, quorumConfig ring.DoUntilQuorumConfig, f func(context.Context, logproto.QuerierClient) (interface{}, error)) ([]responseFromIngesters, error) {
+	results, err := ring.DoUntilQuorum(ctx, replicationSet, quorumConfig, func(ctx context.Context, ingester *ring.InstanceDesc) (responseFromIngesters, error) {
 		client, err := q.pool.GetClientFor(ingester.Addr)
 		if err != nil {
 			return responseFromIngesters{addr: ingester.Addr}, err
@@ -212,7 +249,7 @@ func (q *IngesterQuerier) TailDisconnectedIngesters(ctx context.Context, req *lo
 	}
 
 	// Instance a tail client for each ingester to re(connect)
-	reconnectClients, err := q.forGivenIngesters(ctx, ring.ReplicationSet{Instances: reconnectIngesters}, func(_ context.Context, client logproto.QuerierClient) (interface{}, error) {
+	reconnectClients, err := q.forGivenIngesters(ctx, ring.ReplicationSet{Instances: reconnectIngesters}, defaultQuorumConfig(), func(_ context.Context, client logproto.QuerierClient) (interface{}, error) {
 		return client.Tail(ctx, req)
 	})
 	if err != nil {
@@ -260,7 +297,7 @@ func (q *IngesterQuerier) TailersCount(ctx context.Context) ([]uint32, error) {
 		return nil, httpgrpc.Errorf(http.StatusInternalServerError, "no active ingester found")
 	}
 
-	responses, err := q.forGivenIngesters(ctx, replicationSet, func(ctx context.Context, querierClient logproto.QuerierClient) (interface{}, error) {
+	responses, err := q.forGivenIngesters(ctx, replicationSet, defaultQuorumConfig(), func(ctx context.Context, querierClient logproto.QuerierClient) (interface{}, error) {
 		resp, err := querierClient.TailersCount(ctx, &logproto.TailersCountRequest{})
 		if err != nil {
 			return nil, err

pkg/querier/ingester_querier_test.go

@@ -8,6 +8,7 @@ import (
 	"time"
 
 	"github.com/go-kit/log"
+	"github.com/grafana/dskit/user"
 	"go.uber.org/atomic"
 
 	"google.golang.org/grpc/codes"
@@ -224,6 +225,129 @@ func TestIngesterQuerier_earlyExitOnQuorum(t *testing.T) {
 	}
 }
 
+func TestIngesterQuerierFetchesResponsesFromPartitionIngesters(t *testing.T) {
+	t.Parallel()
+	ctx := user.InjectOrgID(context.Background(), "test-user")
+	ctx, cancel := context.WithTimeout(ctx, time.Second*10)
+	defer cancel()
+
+	ingesters := []ring.InstanceDesc{
+		mockInstanceDescWithZone("1.1.1.1", ring.ACTIVE, "A"),
+		mockInstanceDescWithZone("2.2.2.2", ring.ACTIVE, "B"),
+		mockInstanceDescWithZone("3.3.3.3", ring.ACTIVE, "A"),
+		mockInstanceDescWithZone("4.4.4.4", ring.ACTIVE, "B"),
+		mockInstanceDescWithZone("5.5.5.5", ring.ACTIVE, "A"),
+		mockInstanceDescWithZone("6.6.6.6", ring.ACTIVE, "B"),
+	}
+
+	tests := map[string]struct {
+		method string
+		testFn func(*IngesterQuerier) error
+		retVal interface{}
+		shards int
+	}{
+		"label": {
+			method: "Label",
+			testFn: func(ingesterQuerier *IngesterQuerier) error {
+				_, err := ingesterQuerier.Label(ctx, nil)
+				return err
+			},
+			retVal: new(logproto.LabelResponse),
+		},
+		"series": {
+			method: "Series",
+			testFn: func(ingesterQuerier *IngesterQuerier) error {
+				_, err := ingesterQuerier.Series(ctx, nil)
+				return err
+			},
+			retVal: new(logproto.SeriesResponse),
+		},
+		"get_chunk_ids": {
+			method: "GetChunkIDs",
+			testFn: func(ingesterQuerier *IngesterQuerier) error {
+				_, err := ingesterQuerier.GetChunkIDs(ctx, model.Time(0), model.Time(0))
+				return err
+			},
+			retVal: new(logproto.GetChunkIDsResponse),
+		},
+		"select_logs": {
+			method: "Query",
+			testFn: func(ingesterQuerier *IngesterQuerier) error {
+				_, err := ingesterQuerier.SelectLogs(ctx, logql.SelectLogParams{
+					QueryRequest: new(logproto.QueryRequest),
+				})
+				return err
+			},
+			retVal: newQueryClientMock(),
+		},
+		"select_sample": {
+			method: "QuerySample",
+			testFn: func(ingesterQuerier *IngesterQuerier) error {
+				_, err := ingesterQuerier.SelectSample(ctx, logql.SelectSampleParams{
+					SampleQueryRequest: new(logproto.SampleQueryRequest),
+				})
+				return err
+			},
+			retVal: newQuerySampleClientMock(),
+		},
+		"select_logs_shuffle_sharded": {
+			method: "Query",
+			testFn: func(ingesterQuerier *IngesterQuerier) error {
+				_, err := ingesterQuerier.SelectLogs(ctx, logql.SelectLogParams{
+					QueryRequest: new(logproto.QueryRequest),
+				})
+				return err
+			},
+			retVal: newQueryClientMock(),
+			shards: 2, // Must be less than number of partitions
+		},
+	}
+
+	for testName, testData := range tests {
+		cnt := atomic.NewInt32(0)
+
+		t.Run(testName, func(t *testing.T) {
+			cnt.Store(0)
+			runFn := func(args mock.Arguments) {
+				ctx := args[0].(context.Context)
+
+				select {
+				case <-ctx.Done():
+					// should not be cancelled by the tracker
+					require.NoError(t, ctx.Err())
+				default:
+					cnt.Add(1)
+				}
+			}
+
+			instanceRing := newReadRingMock(ingesters, 0)
+			ingesterClient := newQuerierClientMock()
+			ingesterClient.On(testData.method, mock.Anything, mock.Anything, mock.Anything).Return(testData.retVal, nil).Run(runFn)
+
+			partitions := 3
+			ingestersPerPartition := len(ingesters) / partitions
+			assert.Greaterf(t, ingestersPerPartition, 1, "must have more than one ingester per partition")
+
+			ingesterQuerier, err := newTestPartitionIngesterQuerier(ingesterClient, instanceRing, newPartitionInstanceRingMock(instanceRing, ingesters, partitions, ingestersPerPartition), testData.shards)
+			require.NoError(t, err)
+
+			ingesterQuerier.querierConfig.QueryPartitionIngesters = true
+
+			err = testData.testFn(ingesterQuerier)
+			require.NoError(t, err)
+
+			if testData.shards == 0 {
+				testData.shards = partitions
+			}
+			expectedCalls := min(testData.shards, partitions)
+			// Wait for responses: We expect one request per queried partition because we have request minimization enabled & ingesters are in multiple zones.
+			// If shuffle sharding is enabled, we expect one query per shard as we write to a subset of partitions.
+			require.Eventually(t, func() bool { return cnt.Load() >= int32(expectedCalls) }, time.Millisecond*100, time.Millisecond*1, "expected all ingesters to respond")
+			ingesterClient.AssertNumberOfCalls(t, testData.method, expectedCalls)
+		})
+	}
+}
+
 func TestQuerier_tailDisconnectedIngesters(t *testing.T) {
 	t.Parallel()
 
@@ -400,9 +524,24 @@ func TestIngesterQuerier_DetectedLabels(t *testing.T) {
 
 func newTestIngesterQuerier(readRingMock *readRingMock, ingesterClient *querierClientMock) (*IngesterQuerier, error) {
 	return newIngesterQuerier(
+		mockQuerierConfig(),
 		mockIngesterClientConfig(),
 		readRingMock,
-		mockQuerierConfig().ExtraQueryDelay,
+		nil,
+		func(string) int { return 0 },
+		newIngesterClientMockFactory(ingesterClient),
+		constants.Loki,
+		log.NewNopLogger(),
+	)
+}
+
+func newTestPartitionIngesterQuerier(ingesterClient *querierClientMock, instanceRing *readRingMock, partitionRing *ring.PartitionInstanceRing, tenantShards int) (*IngesterQuerier, error) {
+	return newIngesterQuerier(
+		mockQuerierConfig(),
+		mockIngesterClientConfig(),
+		instanceRing,
+		partitionRing,
+		func(string) int { return tenantShards },
 		newIngesterClientMockFactory(ingesterClient),
 		constants.Loki,
 		log.NewNopLogger(),

pkg/querier/querier.go

@@ -72,6 +72,7 @@ type Config struct {
 	QueryIngesterOnly             bool             `yaml:"query_ingester_only"`
 	MultiTenantQueriesEnabled     bool             `yaml:"multi_tenant_queries_enabled"`
 	PerRequestLimitsEnabled       bool             `yaml:"per_request_limits_enabled"`
+	QueryPartitionIngesters       bool             `yaml:"query_partition_ingesters" category:"experimental"`
 }
 
 // RegisterFlags register flags.
@@ -85,6 +86,7 @@ func (cfg *Config) RegisterFlags(f *flag.FlagSet) {
 	f.BoolVar(&cfg.QueryIngesterOnly, "querier.query-ingester-only", false, "When true, queriers only query the ingesters, and not stored data. This is useful when the object store is unavailable.")
 	f.BoolVar(&cfg.MultiTenantQueriesEnabled, "querier.multi-tenant-queries-enabled", false, "When true, allow queries to span multiple tenants.")
 	f.BoolVar(&cfg.PerRequestLimitsEnabled, "querier.per-request-limits-enabled", false, "When true, querier limits sent via a header are enforced.")
+	f.BoolVar(&cfg.QueryPartitionIngesters, "querier.query_partition_ingesters", false, "When true, querier directs ingester queries to the partition-ingesters instead of the normal ingesters.")
 }
 
 // Validate validates the config.