Read path for RF1 generator blocks [3 of 3] (#3723)

* TraceQL metrics - RF1 read path * Maintain old metrics read path * Fix req time alignment * Address review comments * Update docs * Fix module optionalstore check, it was inverted --------- Co-authored-by: Martin Disibio <[email protected]>
grafana · May 31, 2024 · b4514b9 · b4514b9
1 parent f9bd4fc
commit b4514b9
Show file tree

Hide file tree

Showing 18 changed files with 861 additions and 224 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,7 +1,7 @@
 ## main / unreleased
 
 * [FEATURE] TraceQL support for event scope and event:name intrinsic [#3708](https://github.com/grafana/tempo/pull/3708) (@stoewer)
-* [FEATURE] Flush blocks to storage from the metrics-generator [#3628](https://github.com/grafana/tempo/pull/3628) [#3691](https://github.com/grafana/tempo/pull/3691) (@mapno)
+* [FEATURE] Flush and query RF1 blocks for TraceQL metric queries [#3628](https://github.com/grafana/tempo/pull/3628) [#3691](https://github.com/grafana/tempo/pull/3691) [#3723](https://github.com/grafana/tempo/pull/3723) (@mapno)
 * [ENHANCEMENT] Tag value lookup use protobuf internally for improved latency [#3731](https://github.com/grafana/tempo/pull/3731) (@mdisibio)
 * [ENHANCEMENT] Improve use of OTEL semantic conventions on the service graph [#3711](https://github.com/grafana/tempo/pull/3711) (@zalegrala)
 * [ENHANCEMENT] Performance improvement for `rate() by ()` queries [#3719](https://github.com/grafana/tempo/pull/3719) (@mapno)

diff --git a/cmd/tempo/app/modules.go b/cmd/tempo/app/modules.go
@@ -263,9 +263,8 @@ func (t *App) initIngester() (services.Service, error) {
 }
 
 func (t *App) initGenerator() (services.Service, error) {
-	if t.cfg.Target == MetricsGenerator &&
-		t.cfg.StorageConfig.Trace.Backend != "" &&
-		t.cfg.Generator.Processor.LocalBlocks.FlushToStorage {
+	if t.cfg.Generator.Processor.LocalBlocks.FlushToStorage &&
+		t.store == nil {
 		return nil, fmt.Errorf("generator.processor.local-blocks.flush-to-storage is enabled but no storage backend is configured")
 	}
 

diff --git a/docs/sources/tempo/configuration/_index.md b/docs/sources/tempo/configuration/_index.md
@@ -553,6 +553,9 @@ query_frontend:
         # If set to a non-zero value, it's value will be used to decide if query is within SLO or not.
         # Query is within SLO if it returned 200 within duration_slo seconds OR processed throughput_slo bytes/s data.
         [throughput_bytes_slo: <float> | default = 0 ]
+
+        # If set to true, TraceQL metric queries will use RF1 blocks built and flushed by the metrics-generator.
+        [rf1_read_path: <bool> | default = false]
 ```
 
 ## Querier

diff --git a/docs/sources/tempo/configuration/manifest.md b/docs/sources/tempo/configuration/manifest.md
@@ -74,6 +74,7 @@ server:
     log_format: logfmt
     log_level: info
     log_source_ips_enabled: false
+    log_source_ips_full: false
     log_source_ips_header: ""
     log_source_ips_regex: ""
     log_request_headers: false
@@ -129,6 +130,7 @@ internal_server:
     log_format: logfmt
     log_level: info
     log_source_ips_enabled: false
+    log_source_ips_full: false
     log_source_ips_header: ""
     log_source_ips_regex: ""
     log_request_headers: false
@@ -254,6 +256,9 @@ querier:
         external_endpoints: []
     trace_by_id:
         query_timeout: 10s
+    metrics:
+        concurrent_blocks: 2
+        time_overlap_cutoff: 0.2
     max_concurrent_queries: 20
     frontend_worker:
         frontend_address: 127.0.0.1:9095
@@ -285,11 +290,14 @@ querier:
             connect_timeout: 0s
             connect_backoff_base_delay: 0s
             connect_backoff_max_delay: 0s
+    shuffle_sharding_ingesters_enabled: false
+    shuffle_sharding_ingesters_lookback_period: 1h0m0s
     query_relevant_ingesters: false
 query_frontend:
     max_outstanding_per_tenant: 2000
     querier_forget_delay: 0s
     max_batch_size: 5
+    log_query_request_headers: ""
     max_retries: 2
     search:
         concurrent_jobs: 1000
@@ -299,13 +307,14 @@ query_frontend:
         max_duration: 168h0m0s
         query_backend_after: 15m0s
         query_ingesters_until: 30m0s
+        ingester_shards: 1
     trace_by_id:
         query_shards: 50
     metrics:
         concurrent_jobs: 1000
         target_bytes_per_job: 104857600
-        max_duration: 0s
-        query_backend_after: 1h0m0s
+        max_duration: 3h0m0s
+        query_backend_after: 30m0s
         interval: 5m0s
     multi_tenant_queries_enabled: true
 compactor:
@@ -410,7 +419,6 @@ ingester:
         min_ready_duration: 15s
         interface_names:
             - en0
-            - bridge100
         enable_inet6: false
         final_sleep: 0s
         tokens_file_path: ""
@@ -492,6 +500,7 @@ metrics_generator:
                 - db.name
                 - db.system
             span_multiplier_key: ""
+            enable_virtual_node_label: false
         span_metrics:
             histogram_buckets:
                 - 0.002
@@ -550,8 +559,10 @@ metrics_generator:
             max_block_bytes: 500000000
             complete_block_timeout: 1h0m0s
             max_live_traces: 0
-            concurrent_blocks: 10
             filter_server_spans: true
+            flush_to_storage: false
+            concurrent_blocks: 10
+            time_overlap_cutoff: 0.2
     registry:
         collection_interval: 15s
         stale_duration: 15m0s
@@ -620,6 +631,8 @@ storage:
         blocklist_poll_stale_tenant_index: 0s
         blocklist_poll_jitter_ms: 0
         blocklist_poll_tolerate_consecutive_errors: 1
+        empty_tenant_deletion_enabled: false
+        empty_tenant_deletion_age: 0s
         backend: local
         local:
             path: /var/tempo/traces

diff --git a/modules/frontend/config.go b/modules/frontend/config.go
@@ -85,6 +85,7 @@ func (cfg *Config) RegisterFlagsAndApplyDefaults(string, *flag.FlagSet) {
 			ConcurrentRequests:    defaultConcurrentRequests,
 			TargetBytesPerRequest: defaultTargetBytesPerRequest,
 			Interval:              5 * time.Minute,
+			RF1ReadPath:           false,
 		},
 		SLO: slo,
 	}

diff --git a/modules/frontend/metrics_query_range_sharder.go b/modules/frontend/metrics_query_range_sharder.go
@@ -25,11 +25,12 @@ import (
 )
 
 type queryRangeSharder struct {
-	next      pipeline.AsyncRoundTripper[combiner.PipelineResponse]
-	reader    tempodb.Reader
-	overrides overrides.Interface
-	cfg       QueryRangeSharderConfig
-	logger    log.Logger
+	next              pipeline.AsyncRoundTripper[combiner.PipelineResponse]
+	reader            tempodb.Reader
+	overrides         overrides.Interface
+	cfg               QueryRangeSharderConfig
+	logger            log.Logger
+	replicationFactor uint32
 }
 
 type QueryRangeSharderConfig struct {
@@ -38,10 +39,15 @@ type QueryRangeSharderConfig struct {
 	MaxDuration           time.Duration `yaml:"max_duration"`
 	QueryBackendAfter     time.Duration `yaml:"query_backend_after,omitempty"`
 	Interval              time.Duration `yaml:"interval,omitempty"`
+	RF1ReadPath           bool          `yaml:"rf1_read_path,omitempty"`
 }
 
 // newAsyncQueryRangeSharder creates a sharding middleware for search
 func newAsyncQueryRangeSharder(reader tempodb.Reader, o overrides.Interface, cfg QueryRangeSharderConfig, logger log.Logger) pipeline.AsyncMiddleware[combiner.PipelineResponse] {
+	var replicationFactor uint32
+	if cfg.RF1ReadPath {
+		replicationFactor = 1
+	}
 	return pipeline.AsyncMiddlewareFunc[combiner.PipelineResponse](func(next pipeline.AsyncRoundTripper[combiner.PipelineResponse]) pipeline.AsyncRoundTripper[combiner.PipelineResponse] {
 		return queryRangeSharder{
 			next:      next,
@@ -50,6 +56,8 @@ func newAsyncQueryRangeSharder(reader tempodb.Reader, o overrides.Interface, cfg
 
 			cfg:    cfg,
 			logger: logger,
+
+			replicationFactor: replicationFactor,
 		}
 	})
 }
@@ -110,7 +118,19 @@ func (s queryRangeSharder) RoundTrip(r *http.Request) (pipeline.Responses[combin
 		reqCh <- generatorReq
 	}
 
-	totalJobs, totalBlocks, totalBlockBytes := s.backendRequests(ctx, tenantID, r, *req, now, samplingRate, targetBytesPerRequest, interval, reqCh, nil)
+	var (
+		totalJobs, totalBlocks uint32
+		totalBlockBytes        uint64
+	)
+	if s.cfg.RF1ReadPath {
+		totalJobs, totalBlocks, totalBlockBytes = s.backendRequests(ctx, tenantID, r, *req, now, samplingRate, targetBytesPerRequest, interval, reqCh)
+	} else {
+		totalJobs, totalBlocks, totalBlockBytes = s.shardedBackendRequests(ctx, tenantID, r, *req, now, samplingRate, targetBytesPerRequest, interval, reqCh, nil)
+	}
+
+	span.SetTag("totalJobs", totalJobs)
+	span.SetTag("totalBlocks", totalBlocks)
+	span.SetTag("totalBlockBytes", totalBlockBytes)
 
 	// send a job to communicate the search metrics. this is consumed by the combiner to calculate totalblocks/bytes/jobs
 	var jobMetricsResponse pipeline.Responses[combiner.PipelineResponse]
@@ -142,15 +162,16 @@ func (s *queryRangeSharder) blockMetas(start, end int64, tenantID string) []*bac
 	metas := make([]*backend.BlockMeta, 0, len(allMetas)/50) // divide by 50 for luck
 	for _, m := range allMetas {
 		if m.StartTime.UnixNano() <= end &&
-			m.EndTime.UnixNano() >= start {
+			m.EndTime.UnixNano() >= start &&
+			m.ReplicationFactor == s.replicationFactor {
 			metas = append(metas, m)
 		}
 	}
 
 	return metas
 }
 
-func (s *queryRangeSharder) backendRequests(ctx context.Context, tenantID string, parent *http.Request, searchReq tempopb.QueryRangeRequest, now time.Time, samplingRate float64, targetBytesPerRequest int, interval time.Duration, reqCh chan *http.Request, _ func(error)) (totalJobs, totalBlocks uint32, totalBlockBytes uint64) {
+func (s *queryRangeSharder) shardedBackendRequests(ctx context.Context, tenantID string, parent *http.Request, searchReq tempopb.QueryRangeRequest, now time.Time, samplingRate float64, targetBytesPerRequest int, interval time.Duration, reqCh chan *http.Request, _ func(error)) (totalJobs, totalBlocks uint32, totalBlockBytes uint64) {
 	// request without start or end, search only in generator
 	if searchReq.Start == 0 || searchReq.End == 0 {
 		close(reqCh)
@@ -218,13 +239,13 @@ func (s *queryRangeSharder) backendRequests(ctx context.Context, tenantID string
 	}
 
 	go func() {
-		s.buildBackendRequests(ctx, tenantID, parent, backendReq, samplingRate, targetBytesPerRequest, interval, reqCh, nil)
+		s.buildShardedBackendRequests(ctx, tenantID, parent, backendReq, samplingRate, targetBytesPerRequest, interval, reqCh)
 	}()
 
 	return
 }
 
-func (s *queryRangeSharder) buildBackendRequests(ctx context.Context, tenantID string, parent *http.Request, searchReq tempopb.QueryRangeRequest, samplingRate float64, targetBytesPerRequest int, interval time.Duration, reqCh chan *http.Request, _ func(error)) {
+func (s *queryRangeSharder) buildShardedBackendRequests(ctx context.Context, tenantID string, parent *http.Request, searchReq tempopb.QueryRangeRequest, samplingRate float64, targetBytesPerRequest int, interval time.Duration, reqCh chan *http.Request) {
 	defer close(reqCh)
 
 	var (
@@ -260,7 +281,6 @@ func (s *queryRangeSharder) buildBackendRequests(ctx context.Context, tenantID s
 			shardR.End = thisEnd
 			shardR.ShardID = i
 			shardR.ShardCount = shards
-
 			httpReq := s.toUpstreamRequest(ctx, shardR, parent, tenantID)
 			if samplingRate != 1.0 {
 				shardR.ShardID *= uint32(1.0 / samplingRate)
@@ -283,6 +303,105 @@ func (s *queryRangeSharder) buildBackendRequests(ctx context.Context, tenantID s
 	}
 }
 
+func (s *queryRangeSharder) backendRequests(ctx context.Context, tenantID string, parent *http.Request, searchReq tempopb.QueryRangeRequest, now time.Time, _ float64, targetBytesPerRequest int, _ time.Duration, reqCh chan *http.Request) (totalJobs, totalBlocks uint32, totalBlockBytes uint64) {
+	// request without start or end, search only in generator
+	if searchReq.Start == 0 || searchReq.End == 0 {
+		close(reqCh)
+		return
+	}
+
+	// Make a copy and limit to backend time range.
+	backendReq := searchReq
+	backendReq.Start, backendReq.End = s.backendRange(now, backendReq.Start, backendReq.End, s.cfg.QueryBackendAfter)
+	alignTimeRange(&backendReq)
+
+	// If empty window then no need to search backend
+	if backendReq.Start == backendReq.End {
+		close(reqCh)
+		return
+	}
+
+	// Blocks within overall time range. This is just for instrumentation, more precise time
+	// range is checked for each window.
+	blocks := s.blockMetas(int64(backendReq.Start), int64(backendReq.End), tenantID)
+	if len(blocks) == 0 {
+		// no need to search backend
+		close(reqCh)
+		return
+	}
+
+	// calculate metrics to return to the caller
+	totalBlocks = uint32(len(blocks))
+	for _, b := range blocks {
+		p := pagesPerRequest(b, targetBytesPerRequest)
+
+		totalJobs += b.TotalRecords / uint32(p)
+		if int(b.TotalRecords)%p != 0 {
+			totalJobs++
+		}
+		totalBlockBytes += b.Size
+	}
+
+	go func() {
+		s.buildBackendRequests(ctx, tenantID, parent, backendReq, blocks, targetBytesPerRequest, reqCh)
+	}()
+
+	return
+}
+
+func (s *queryRangeSharder) buildBackendRequests(ctx context.Context, tenantID string, parent *http.Request, searchReq tempopb.QueryRangeRequest, metas []*backend.BlockMeta, targetBytesPerRequest int, reqCh chan<- *http.Request) {
+	defer close(reqCh)
+
+	for _, m := range metas {
+		pages := pagesPerRequest(m, targetBytesPerRequest)
+		if pages == 0 {
+			continue
+		}
+
+		for startPage := 0; startPage < int(m.TotalRecords); startPage += pages {
+			subR := parent.Clone(ctx)
+
+			dc, err := m.DedicatedColumns.ToTempopb()
+			if err != nil {
+				// errFn(fmt.Errorf("failed to convert dedicated columns. block: %s tempopb: %w", blockID, err))
+				continue
+			}
+
+			queryRangeReq := &tempopb.QueryRangeRequest{
+				Query: searchReq.Query,
+				Start: max(searchReq.Start, uint64(m.StartTime.UnixNano())),
+				End:   min(searchReq.End, uint64(m.EndTime.UnixNano())),
+				Step:  searchReq.Step,
+				// ShardID:    uint32, // No sharding with RF=1
+				// ShardCount: uint32, // No sharding with RF=1
+				QueryMode: searchReq.QueryMode,
+				// New RF1 fields
+				BlockID:          m.BlockID.String(),
+				StartPage:        uint32(startPage),
+				PagesToSearch:    uint32(pages),
+				Version:          m.Version,
+				Encoding:         m.Encoding.String(),
+				Size_:            m.Size,
+				FooterSize:       m.FooterSize,
+				DedicatedColumns: dc,
+			}
+			alignTimeRange(queryRangeReq)
+			queryRangeReq.End += queryRangeReq.Step
+
+			subR = api.BuildQueryRangeRequest(subR, queryRangeReq)
+
+			prepareRequestForQueriers(subR, tenantID, subR.URL.Path, subR.URL.Query())
+			// TODO: Handle sampling rate
+
+			select {
+			case reqCh <- subR:
+			case <-ctx.Done():
+				return
+			}
+		}
+	}
+}
+
 func (s *queryRangeSharder) backendRange(now time.Time, start, end uint64, queryBackendAfter time.Duration) (uint64, uint64) {
 	backendAfter := uint64(now.Add(-queryBackendAfter).UnixNano())
 
@@ -319,13 +438,6 @@ func (s *queryRangeSharder) generatorRequest(searchReq tempopb.QueryRangeRequest
 
 	searchReq.QueryMode = querier.QueryModeRecent
 
-	// No sharding on the generators (unnecessary), but we do apply sampling
-	// rates.  In this case we execute a single arbitrary shard. Choosing
-	// the last shard works. The first shard should be avoided because it is
-	// weighted slightly off due to int63/128 sharding boundaries.
-	searchReq.ShardID = uint32(1.0 / samplingRate)
-	searchReq.ShardCount = uint32(1.0 / samplingRate)
-
 	return s.toUpstreamRequest(parent.Context(), searchReq, parent, tenantID)
 }
 

diff --git a/modules/frontend/search_sharder.go b/modules/frontend/search_sharder.go
@@ -38,11 +38,6 @@ type SearchSharderConfig struct {
 	IngesterShards        int           `yaml:"ingester_shards,omitempty"`
 }
 
-type backendReqMsg struct {
-	req *http.Request
-	err error
-}
-
 type asyncSearchSharder struct {
 	next      pipeline.AsyncRoundTripper[combiner.PipelineResponse]
 	reader    tempodb.Reader
@@ -147,7 +142,8 @@ func (s *asyncSearchSharder) blockMetas(start, end int64, tenantID string) []*ba
 	metas := make([]*backend.BlockMeta, 0, len(allMetas)/50) // divide by 50 for luck
 	for _, m := range allMetas {
 		if m.StartTime.Unix() <= end &&
-			m.EndTime.Unix() >= start {
+			m.EndTime.Unix() >= start &&
+			m.ReplicationFactor == backend.DefaultReplicationFactor { // This check skips generator blocks (RF=1)
 			metas = append(metas, m)
 		}
 	}

diff --git a/modules/generator/processor/localblocks/processor.go b/modules/generator/processor/localblocks/processor.go
@@ -728,7 +728,7 @@ func (p *Processor) resetHeadBlock() error {
 		BlockID:           uuid.New(),
 		TenantID:          p.tenant,
 		DedicatedColumns:  p.overrides.DedicatedColumns(p.tenant),
-		ReplicationFactor: 1,
+		ReplicationFactor: backend.MetricsGeneratorReplicationFactor,
 	}
 	block, err := p.wal.NewBlock(meta, model.CurrentEncoding)
 	if err != nil {

diff --git a/modules/querier/querier.go b/modules/querier/querier.go
@@ -284,6 +284,7 @@ func (q *Querier) FindTraceByID(ctx context.Context, req *tempopb.TraceByIDReque
 		span.LogFields(ot_log.String("timeEnd", fmt.Sprint(timeEnd)))
 
 		opts := common.DefaultSearchOptionsWithMaxBytes(maxBytes)
+		opts.BlockReplicationFactor = backend.DefaultReplicationFactor
 		partialTraces, blockErrs, err := q.store.Find(ctx, userID, req.TraceID, req.BlockStart, req.BlockEnd, timeStart, timeEnd, opts)
 		if err != nil {
 			retErr := fmt.Errorf("error querying store in Querier.FindTraceByID: %w", err)