fix: do not add system MID to fraction distribution (#256)

Author: cheb0

frac/active_sealing_source.go

@@ -147,7 +147,12 @@ func (src *ActiveSealingSource) LastError() error {
 func (src *ActiveSealingSource) prepareInfo() {
 	src.info.MetaOnDisk = 0
 	src.info.SealingTime = uint64(src.created.UnixMilli())
-	src.info.BuildDistribution(src.mids.vals)
+	mids := src.mids.vals
+	if len(mids) > 1 {
+		// skip system MID
+		mids = mids[1:]
+	}
+	src.info.BuildDistribution(mids)
 }
 
 // Info returns index metadata information.

frac/fraction_test.go

@@ -1026,6 +1026,33 @@ func (s *FractionTestSuite) TestSearchLargeFrac() {
 	s.AssertSearch(s.query("level:5", withLimit(100)), docs, level5Indexes[:100])
 }
 
+func (s *FractionTestSuite) TestMIDDistribution() {
+	now := time.Now().Truncate(time.Minute)
+	docs := []string{
+		fmt.Sprintf(`{"timestamp":%q,"message":"apple juice"}`, now.Add(-60*time.Minute).Format(time.RFC3339Nano)),
+		fmt.Sprintf(`{"timestamp":%q,"message":"orange juice"}`, now.Add(-61*time.Minute).Format(time.RFC3339Nano)),
+		fmt.Sprintf(`{"timestamp":%q,"message":"cider"}`, now.Add(-65*time.Minute).Format(time.RFC3339Nano)),
+		fmt.Sprintf(`{"timestamp":%q,"message":"red wine"}`, now.Add(-120*time.Minute).Format(time.RFC3339Nano)),
+		fmt.Sprintf(`{"timestamp":%q,"message":"coca cola"}`, now.Add(-360*time.Minute).Format(time.RFC3339Nano)),
+	}
+
+	s.insertDocuments(docs)
+
+	_, ok := s.fraction.(*Active)
+	if ok {
+		s.Require().Nil(s.fraction.Info().Distribution, "active fraction has MID distribution")
+		return
+	}
+
+	dist := s.fraction.Info().Distribution.GetDist()
+	s.Require().Equal(5, len(dist))
+	s.Require().Equal(now.Add(-360*time.Minute).UTC(), dist[0])
+	s.Require().Equal(now.Add(-120*time.Minute).UTC(), dist[1])
+	s.Require().Equal(now.Add(-65*time.Minute).UTC(), dist[2])
+	s.Require().Equal(now.Add(-61*time.Minute).UTC(), dist[3])
+	s.Require().Equal(now.Add(-60*time.Minute).UTC(), dist[4])
+}
+
 func (s *FractionTestSuite) TestFractionInfo() {
 	docs := []string{
 		`{"timestamp":"2000-01-01T13:00:25Z","service":"service_a","message":"first message some text", "container":"gateway"}`,

seq/seq.go

@@ -4,6 +4,7 @@ import (
 	"encoding/binary"
 	"encoding/hex"
 	"fmt"
+	"math"
 	"time"
 
 	"github.com/ozontech/seq-db/util"
@@ -19,7 +20,12 @@ type RID uint64 // random part of ID
 type LID uint32 // local id for a fraction
 
 func (m MID) Time() time.Time {
-	return time.UnixMilli(int64(m))
+	if uint64(m) <= math.MaxInt64 {
+		return time.UnixMilli(int64(m))
+	} else {
+		// since MaxInt64 is 292278994 year in milliseconds, so we assume this MID is "infinite future"
+		return time.UnixMilli(math.MaxInt64)
+	}
 }
 
 func (d ID) String() string {
@@ -105,7 +111,7 @@ func DurationToMID(d time.Duration) MID {
 }
 
 func MIDToTime(t MID) time.Time {
-	return time.Unix(0, 0).Add(MIDToDuration(t))
+	return t.Time()
 }
 
 func MIDToDuration(t MID) time.Duration {

seq/seq_test.go

@@ -1,7 +1,9 @@
 package seq
 
 import (
+	"math"
 	"testing"
+	"time"
 
 	"github.com/stretchr/testify/assert"
 )
@@ -10,3 +12,22 @@ func TestFromString(t *testing.T) {
 	_, err := FromString("abaf05877b010000-2402dc02d60615cc")
 	assert.NoError(t, err)
 }
+
+func TestTimeToMIDConversion(t *testing.T) {
+	timestampNow := time.Now()
+	assert.EqualExportedValues(t, timestampNow, MID(timestampNow.UnixNano()).Time())
+
+	timestamp2 := MID(1763984556395).Time().UTC()
+	assert.Equal(t, 2025, timestamp2.Year())
+	assert.Equal(t, time.Month(11), timestamp2.Month())
+	assert.Equal(t, 24, timestamp2.Day())
+	assert.Equal(t, 11, timestamp2.Hour())
+	assert.Equal(t, 42, timestamp2.Minute())
+	assert.Equal(t, 36, timestamp2.Second())
+	assert.Equal(t, 395000000, timestamp2.Nanosecond())
+
+	// check that we do not overflow on huge values
+	maxMID := MID(math.MaxUint64)
+	assert.Equal(t, 292278994, maxMID.Time().Year())
+	assert.Equal(t, 292278994, MIDToTime(maxMID).Year())
+}