Add per-user query metrics for series and bytes returned (cortexproject#4343)

* Add per-user query metrics for series and bytes returned

Add stats included in query responses from the querier and distributor
for measuring the number of series and bytes included in successful
queries. These stats are emitted per-user as summaries from the query
frontends.

These stats are picked to add visibility into the same resources limited
as part of cortexproject#4179 and cortexproject#4216.

Fixes cortexproject#4259

Signed-off-by: Nick Pillitteri <[email protected]>

* Formatting fix

Signed-off-by: Nick Pillitteri <[email protected]>

* Fix changelog to match actual changes

Signed-off-by: Nick Pillitteri <[email protected]>

* Typo

Signed-off-by: Nick Pillitteri <[email protected]>

* Code review changes, rename things for clarity

Signed-off-by: Nick Pillitteri <[email protected]>

* Apply suggestions from code review

Co-authored-by: Marco Pracucci <[email protected]>
Signed-off-by: Nick Pillitteri <[email protected]>

* Code review changes, remove superfluous summaries

Signed-off-by: Nick Pillitteri <[email protected]>

Co-authored-by: Marco Pracucci <[email protected]>
Signed-off-by: Alvin Lin <[email protected]>

Author: 2 people

CHANGELOG.md

@@ -3,6 +3,7 @@
 ## master / unreleased
 * [FEATURE] Ruler: Add new `-ruler.query-stats-enabled` which when enabled will report the `cortex_ruler_query_seconds_total` as a per-user metric that tracks the sum of the wall time of executing queries in the ruler in seconds. #4317
 * [FEATURE] Add shuffle sharding grouper and planner within compactor to allow further work towards parallelizing compaction #4357
+* [FEATURE] Query Frontend: Add `cortex_query_fetched_series_total` and `cortex_query_fetched_chunks_bytes_total` per-user counters to expose the number of series and bytes fetched as part of queries. These metrics can be enabled with the `-frontend.query-stats-enabled` flag (or its respective YAML config option `query_stats_enabled`). #4343
 * [CHANGE] Update Go version to 1.16.6. #4362
 * [CHANGE] Querier / ruler: Change `-querier.max-fetched-chunks-per-query` configuration to limit to maximum number of chunks that can be fetched in a single query. The number of chunks fetched by ingesters AND long-term storare combined should not exceed the value configured on `-querier.max-fetched-chunks-per-query`. #4260
 * [CHANGE] Memberlist: the `memberlist_kv_store_value_bytes` has been removed due to values no longer being stored in-memory as encoded bytes. #4345

pkg/distributor/query.go

@@ -13,6 +13,7 @@ import (
 
 	"github.com/cortexproject/cortex/pkg/cortexpb"
 	ingester_client "github.com/cortexproject/cortex/pkg/ingester/client"
+	"github.com/cortexproject/cortex/pkg/querier/stats"
 	"github.com/cortexproject/cortex/pkg/ring"
 	"github.com/cortexproject/cortex/pkg/tenant"
 	"github.com/cortexproject/cortex/pkg/util"
@@ -282,6 +283,7 @@ func (d *Distributor) queryIngestersExemplars(ctx context.Context, replicationSe
 func (d *Distributor) queryIngesterStream(ctx context.Context, replicationSet ring.ReplicationSet, req *ingester_client.QueryRequest) (*ingester_client.QueryStreamResponse, error) {
 	var (
 		queryLimiter = limiter.QueryLimiterFromContextWithFallback(ctx)
+		reqStats     = stats.FromContext(ctx)
 	)
 
 	// Fetch samples from multiple ingesters
@@ -383,6 +385,9 @@ func (d *Distributor) queryIngesterStream(ctx context.Context, replicationSet ri
 		resp.Timeseries = append(resp.Timeseries, series)
 	}
 
+	reqStats.AddFetchedSeries(uint64(len(resp.Chunkseries) + len(resp.Timeseries)))
+	reqStats.AddFetchedChunkBytes(uint64(resp.ChunksSize()))
+
 	return resp, nil
 }
 

pkg/frontend/transport/handler.go

@@ -60,6 +60,8 @@ type Handler struct {
 
 	// Metrics.
 	querySeconds *prometheus.CounterVec
+	querySeries  *prometheus.CounterVec
+	queryBytes   *prometheus.CounterVec
 	activeUsers  *util.ActiveUsersCleanupService
 }
 
@@ -77,8 +79,20 @@ func NewHandler(cfg HandlerConfig, roundTripper http.RoundTripper, log log.Logge
 			Help: "Total amount of wall clock time spend processing queries.",
 		}, []string{"user"})
 
+		h.querySeries = promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
+			Name: "cortex_query_fetched_series_total",
+			Help: "Number of series fetched to execute a query.",
+		}, []string{"user"})
+
+		h.queryBytes = promauto.With(reg).NewCounterVec(prometheus.CounterOpts{
+			Name: "cortex_query_fetched_chunks_bytes_total",
+			Help: "Size of all chunks fetched to execute a query in bytes.",
+		}, []string{"user"})
+
 		h.activeUsers = util.NewActiveUsersCleanupWithDefaultValues(func(user string) {
 			h.querySeconds.DeleteLabelValues(user)
+			h.querySeries.DeleteLabelValues(user)
+			h.queryBytes.DeleteLabelValues(user)
 		})
 		// If cleaner stops or fail, we will simply not clean the metrics for inactive users.
 		_ = h.activeUsers.StartAsync(context.Background())
@@ -165,9 +179,14 @@ func (f *Handler) reportQueryStats(r *http.Request, queryString url.Values, quer
 		return
 	}
 	userID := tenant.JoinTenantIDs(tenantIDs)
+	wallTime := stats.LoadWallTime()
+	numSeries := stats.LoadFetchedSeries()
+	numBytes := stats.LoadFetchedChunkBytes()
 
 	// Track stats.
-	f.querySeconds.WithLabelValues(userID).Add(stats.LoadWallTime().Seconds())
+	f.querySeconds.WithLabelValues(userID).Add(wallTime.Seconds())
+	f.querySeries.WithLabelValues(userID).Add(float64(numSeries))
+	f.queryBytes.WithLabelValues(userID).Add(float64(numBytes))
 	f.activeUsers.UpdateUserTimestamp(userID, time.Now())
 
 	// Log stats.
@@ -177,7 +196,9 @@ func (f *Handler) reportQueryStats(r *http.Request, queryString url.Values, quer
 		"method", r.Method,
 		"path", r.URL.Path,
 		"response_time", queryResponseTime,
-		"query_wall_time_seconds", stats.LoadWallTime().Seconds(),
+		"query_wall_time_seconds", wallTime.Seconds(),
+		"fetched_series_count", numSeries,
+		"fetched_chunks_bytes", numBytes,
 	}, formatQueryString(queryString)...)
 
 	level.Info(util_log.WithContext(r.Context(), f.log)).Log(logMessage...)

pkg/frontend/transport/handler_test.go

@@ -2,15 +2,28 @@ package transport
 
 import (
 	"context"
+	"io"
 	"net/http"
 	"net/http/httptest"
+	"strings"
 	"testing"
 
+	"github.com/go-kit/kit/log"
 	"github.com/pkg/errors"
+	"github.com/prometheus/client_golang/prometheus"
+	promtest "github.com/prometheus/client_golang/prometheus/testutil"
+	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 	"github.com/weaveworks/common/httpgrpc"
+	"github.com/weaveworks/common/user"
 )
 
+type roundTripperFunc func(*http.Request) (*http.Response, error)
+
+func (f roundTripperFunc) RoundTrip(r *http.Request) (*http.Response, error) {
+	return f(r)
+}
+
 func TestWriteError(t *testing.T) {
 	for _, test := range []struct {
 		status int
@@ -28,3 +41,53 @@ func TestWriteError(t *testing.T) {
 		})
 	}
 }
+
+func TestHandler_ServeHTTP(t *testing.T) {
+	for _, tt := range []struct {
+		name            string
+		cfg             HandlerConfig
+		expectedMetrics int
+	}{
+		{
+			name:            "test handler with stats enabled",
+			cfg:             HandlerConfig{QueryStatsEnabled: true},
+			expectedMetrics: 3,
+		},
+		{
+			name:            "test handler with stats disabled",
+			cfg:             HandlerConfig{QueryStatsEnabled: false},
+			expectedMetrics: 0,
+		},
+	} {
+		t.Run(tt.name, func(t *testing.T) {
+			roundTripper := roundTripperFunc(func(req *http.Request) (*http.Response, error) {
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(strings.NewReader("{}")),
+				}, nil
+			})
+
+			reg := prometheus.NewPedanticRegistry()
+			handler := NewHandler(tt.cfg, roundTripper, log.NewNopLogger(), reg)
+
+			ctx := user.InjectOrgID(context.Background(), "12345")
+			req := httptest.NewRequest("GET", "/", nil)
+			req = req.WithContext(ctx)
+			resp := httptest.NewRecorder()
+
+			handler.ServeHTTP(resp, req)
+			_, _ = io.ReadAll(resp.Body)
+			require.Equal(t, resp.Code, http.StatusOK)
+
+			count, err := promtest.GatherAndCount(
+				reg,
+				"cortex_query_seconds_total",
+				"cortex_query_fetched_series_total",
+				"cortex_query_fetched_chunks_bytes_total",
+			)
+
+			assert.NoError(t, err)
+			assert.Equal(t, tt.expectedMetrics, count)
+		})
+	}
+}

pkg/querier/blocks_store_queryable.go

@@ -29,6 +29,7 @@ import (
 
 	"github.com/cortexproject/cortex/pkg/cortexpb"
 	"github.com/cortexproject/cortex/pkg/querier/series"
+	"github.com/cortexproject/cortex/pkg/querier/stats"
 	"github.com/cortexproject/cortex/pkg/ring"
 	"github.com/cortexproject/cortex/pkg/ring/kv"
 	"github.com/cortexproject/cortex/pkg/storage/bucket"
@@ -565,6 +566,7 @@ func (q *blocksStoreQuerier) fetchSeriesFromStores(
 		numChunks     = atomic.NewInt32(0)
 		spanLog       = spanlogger.FromContext(ctx)
 		queryLimiter  = limiter.QueryLimiterFromContextWithFallback(ctx)
+		reqStats      = stats.FromContext(ctx)
 	)
 
 	// Concurrently fetch series from all clients.
@@ -626,10 +628,7 @@ func (q *blocksStoreQuerier) fetchSeriesFromStores(
 							return validation.LimitError(fmt.Sprintf(errMaxChunksPerQueryLimit, util.LabelMatchersToString(matchers), maxChunksLimit))
 						}
 					}
-					chunksSize := 0
-					for _, c := range s.Chunks {
-						chunksSize += c.Size()
-					}
+					chunksSize := countChunkBytes(s)
 					if chunkBytesLimitErr := queryLimiter.AddChunkBytes(chunksSize); chunkBytesLimitErr != nil {
 						return validation.LimitError(chunkBytesLimitErr.Error())
 					}
@@ -657,10 +656,16 @@ func (q *blocksStoreQuerier) fetchSeriesFromStores(
 				}
 			}
 
+			numSeries := len(mySeries)
+			chunkBytes := countChunkBytes(mySeries...)
+
+			reqStats.AddFetchedSeries(uint64(numSeries))
+			reqStats.AddFetchedChunkBytes(uint64(chunkBytes))
+
 			level.Debug(spanLog).Log("msg", "received series from store-gateway",
 				"instance", c.RemoteAddress(),
-				"num series", len(mySeries),
-				"bytes series", countSeriesBytes(mySeries),
+				"fetched series", numSeries,
+				"fetched chunk bytes", chunkBytes,
 				"requested blocks", strings.Join(convertULIDsToString(blockIDs), " "),
 				"queried blocks", strings.Join(convertULIDsToString(myQueriedBlocks), " "))
 
@@ -944,12 +949,11 @@ func convertBlockHintsToULIDs(hints []hintspb.Block) ([]ulid.ULID, error) {
 	return res, nil
 }
 
-func countSeriesBytes(series []*storepb.Series) (count uint64) {
+// countChunkBytes returns the size of the chunks making up the provided series in bytes
+func countChunkBytes(series ...*storepb.Series) (count int) {
 	for _, s := range series {
 		for _, c := range s.Chunks {
-			if c.Raw != nil {
-				count += uint64(len(c.Raw.Data))
-			}
+			count += c.Size()
 		}
 	}
 

pkg/querier/stats/stats.go

@@ -54,13 +54,47 @@ func (s *Stats) LoadWallTime() time.Duration {
 	return time.Duration(atomic.LoadInt64((*int64)(&s.WallTime)))
 }
 
+func (s *Stats) AddFetchedSeries(series uint64) {
+	if s == nil {
+		return
+	}
+
+	atomic.AddUint64(&s.FetchedSeriesCount, series)
+}
+
+func (s *Stats) LoadFetchedSeries() uint64 {
+	if s == nil {
+		return 0
+	}
+
+	return atomic.LoadUint64(&s.FetchedSeriesCount)
+}
+
+func (s *Stats) AddFetchedChunkBytes(bytes uint64) {
+	if s == nil {
+		return
+	}
+
+	atomic.AddUint64(&s.FetchedChunkBytes, bytes)
+}
+
+func (s *Stats) LoadFetchedChunkBytes() uint64 {
+	if s == nil {
+		return 0
+	}
+
+	return atomic.LoadUint64(&s.FetchedChunkBytes)
+}
+
 // Merge the provide Stats into this one.
 func (s *Stats) Merge(other *Stats) {
 	if s == nil || other == nil {
 		return
 	}
 
 	s.AddWallTime(other.LoadWallTime())
+	s.AddFetchedSeries(other.LoadFetchedSeries())
+	s.AddFetchedChunkBytes(other.LoadFetchedChunkBytes())
 }
 
 func ShouldTrackHTTPGRPCResponse(r *httpgrpc.HTTPResponse) bool {