[PLAT-108920] not skip samples during dedup

pkg/dedup/iter.go

@@ -277,7 +277,7 @@ type dedupSeriesIterator struct {
 	lastIter chunkenc.Iterator
 
 	penA, penB int64
-	useA       bool
+	useA, useB bool
 }
 
 func newDedupSeriesIterator(a, b adjustableSeriesIterator) *dedupSeriesIterator {
@@ -288,6 +288,7 @@ func newDedupSeriesIterator(a, b adjustableSeriesIterator) *dedupSeriesIterator
 		lastIter: a,
 		aval:     a.Next(),
 		bval:     b.Next(),
+		useB:     true,
 	}
 }
 
@@ -304,16 +305,17 @@ func (it *dedupSeriesIterator) Next() chunkenc.ValueType {
 	}()
 
 	// Advance both iterators to at least the next highest timestamp plus the potential penalty.
-	if it.aval != chunkenc.ValNone {
+	if it.aval != chunkenc.ValNone && it.useA {
 		it.aval = it.a.Seek(it.lastT + 1 + it.penA)
 	}
-	if it.bval != chunkenc.ValNone {
+	if it.bval != chunkenc.ValNone && it.useB {
 		it.bval = it.b.Seek(it.lastT + 1 + it.penB)
 	}
 
 	// Handle basic cases where one iterator is exhausted before the other.
 	if it.aval == chunkenc.ValNone {
 		it.useA = false
+		it.useB = true
 		if it.bval != chunkenc.ValNone {
 			it.lastT = it.b.AtT()
 			it.lastIter = it.b
@@ -323,6 +325,7 @@ func (it *dedupSeriesIterator) Next() chunkenc.ValueType {
 	}
 	if it.bval == chunkenc.ValNone {
 		it.useA = true
+		it.useB = false
 		it.lastT = it.a.AtT()
 		it.lastIter = it.a
 		it.penA = 0
@@ -336,6 +339,7 @@ func (it *dedupSeriesIterator) Next() chunkenc.ValueType {
 	tb := it.b.AtT()
 
 	it.useA = ta <= tb
+	it.useB = ta >= tb
 
 	// For the series we didn't pick, add a penalty twice as high as the delta of the last two
 	// samples to the next seek against it.

pkg/dedup/iter_test.go

@@ -97,6 +97,9 @@ func (s *mockedSeriesIterator) Seek(t int64) chunkenc.ValueType {
 }
 
 func (s *mockedSeriesIterator) At() (t int64, v float64) {
+	if s.cur >= len(s.samples) {
+		return 0, 0
+	}
 	sample := s.samples[s.cur]
 	return sample.t, sample.f
 }
@@ -374,6 +377,27 @@ func TestDedupSeriesSet(t *testing.T) {
 				},
 			},
 		},
+		{
+			name: "Multi dedup labels - data points absent",
+			input: []series{
+				{
+					lset:    labels.Labels{{Name: "a", Value: "5"}, {Name: "c", Value: "6"}},
+					samples: []sample{{10000, 1}, {30000, 3}, {40000, 4}},
+				}, {
+					lset:    labels.Labels{{Name: "a", Value: "5"}, {Name: "c", Value: "6"}},
+					samples: []sample{{10000, 1}, {20000, 2}, {30000, 3}, {50000, 5}},
+				}, {
+					lset:    labels.Labels{{Name: "a", Value: "5"}, {Name: "c", Value: "6"}},
+					samples: []sample{{10000, 1}, {80000, 10}},
+				},
+			},
+			exp: []series{
+				{
+					lset:    labels.Labels{{Name: "a", Value: "5"}, {Name: "c", Value: "6"}},
+					samples: []sample{{10000, 1}, {20000, 2}, {30000, 3}, {40000, 4}, {50000, 5}, {80000, 10}},
+				},
+			},
+		},
 		{
 			name: "Multi dedup label - some series don't have all dedup labels",
 			input: []series{

pkg/query/querier.go

@@ -198,10 +198,6 @@ func newQuerierInternal(
 	if logger == nil {
 		logger = log.NewNopLogger()
 	}
-	rl := make(map[string]struct{})
-	for _, replicaLabel := range replicaLabels {
-		rl[replicaLabel] = struct{}{}
-	}
 
 	partialResponseStrategy := storepb.PartialResponseStrategy_ABORT
 	if groupReplicaPartialResponseStrategy {

pkg/query/querier_test.go

@@ -478,7 +478,7 @@ func TestQuerier_Select(t *testing.T) {
 			expectedAfterDedup: []series{{
 				lset: nil,
 				// We don't expect correctness here, it's just random non-replica data.
-				samples: []sample{{1, 1}, {2, 2}, {3, 3}, {5, 5}, {6, 6}, {7, 7}},
+				samples: []sample{{1, 1}, {2, 2}, {3, 3}, {4, 4}, {5, 5}, {6, 6}, {7, 7}},
 			}},
 			expectedWarning: "partial error",
 		},

pkg/receive/multitsdb.go

@@ -590,7 +590,7 @@ func (t *MultiTSDB) startTSDB(logger log.Logger, tenantID string, tenant *tenant
 	opts.EnableOverlappingCompaction = false
 	s, err := tsdb.Open(
 		dataDir,
-		logger,
+		level.NewFilter(logger, level.AllowError()), // warning is too verbose
 		reg,
 		&opts,
 		nil,