Add partitioned group creation time to visit marker

CHANGELOG.md

@@ -35,6 +35,7 @@
 * [ENHANCEMENT] Ingester: Instrument Ingester CPU profile with userID for read APIs. #7184
 * [ENHANCEMENT] Ingester: Add fetch timeout for Ingester expanded postings cache. #7185
 * [ENHANCEMENT] Ingester: Add feature flag to collect metrics of how expensive an unoptimized regex matcher is and new limits to protect Ingester query path against expensive unoptimized regex matchers. #7194 #7210
+* [ENHANCEMENT] Compactor: Add partition group creation time to visit marker. #7217
 * [BUGFIX] Ring: Change DynamoDB KV to retry indefinitely for WatchKey. #7088
 * [BUGFIX] Ruler: Add XFunctions validation support. #7111
 * [BUGFIX] Querier: propagate Prometheus info annotations in protobuf responses. #7132

pkg/compactor/blocks_cleaner_test.go

@@ -1000,9 +1000,9 @@ func TestBlocksCleaner_CleanPartitionedGroupInfo(t *testing.T) {
 	require.NoError(t, err)
 	require.False(t, partitionedGroupFileExists)
 
-	partitionedGroupFileExists, err = userBucket.Exists(ctx, visitMarker.GetVisitMarkerFilePath())
+	visitMarkerExists, err := userBucket.Exists(ctx, visitMarker.GetVisitMarkerFilePath())
 	require.NoError(t, err)
-	require.False(t, partitionedGroupFileExists)
+	require.False(t, visitMarkerExists)
 }
 
 func TestBlocksCleaner_DeleteEmptyBucketIndex(t *testing.T) {

pkg/compactor/partition_compaction_grouper.go

@@ -632,7 +632,7 @@ func (g *PartitionCompactionGrouper) pickPartitionCompactionJob(partitionCompact
 		partitionCount := partitionedGroup.partitionedGroupInfo.PartitionCount
 		partitionID := partitionedGroup.partition.PartitionID
 		partitionedGroupLogger := log.With(g.logger, "rangeStart", partitionedGroup.rangeStartTime().String(), "rangeEnd", partitionedGroup.rangeEndTime().String(), "rangeDuration", partitionedGroup.rangeDuration().String(), "partitioned_group_id", partitionedGroupID, "partition_id", partitionID, "partition_count", partitionCount, "group_hash", groupHash)
-		visitMarker := newPartitionVisitMarker(g.ringLifecyclerID, partitionedGroupID, partitionID)
+		visitMarker := newPartitionVisitMarker(g.ringLifecyclerID, partitionedGroupID, partitionedGroup.partitionedGroupInfo.CreationTime, partitionID)
 		visitMarkerManager := NewVisitMarkerManager(g.bkt, g.logger, g.ringLifecyclerID, visitMarker)
 		if isVisited, err := g.isGroupVisited(partitionID, visitMarkerManager); err != nil {
 			level.Warn(partitionedGroupLogger).Log("msg", "unable to check if partition is visited", "err", err, "group", partitionedGroup.String())

pkg/compactor/partition_compaction_planner.go

@@ -85,7 +85,7 @@ func (p *PartitionCompactionPlanner) PlanWithPartition(_ context.Context, metasB
 	// claimed same partition in grouper at same time.
 	time.Sleep(p.plannerDelay)
 
-	visitMarker := newPartitionVisitMarker(p.ringLifecyclerID, partitionedGroupID, partitionID)
+	visitMarker := newPartitionVisitMarker(p.ringLifecyclerID, partitionedGroupID, partitionInfo.PartitionedGroupCreationTime, partitionID)
 	visitMarkerManager := NewVisitMarkerManager(p.bkt, p.logger, p.ringLifecyclerID, visitMarker)
 	existingPartitionVisitMarker := &partitionVisitMarker{}
 	err := visitMarkerManager.ReadVisitMarker(p.ctx, existingPartitionVisitMarker)

pkg/compactor/partition_visit_marker.go

@@ -28,18 +28,22 @@ type partitionVisitMarker struct {
 	CompactorID        string      `json:"compactorID"`
 	Status             VisitStatus `json:"status"`
 	PartitionedGroupID uint32      `json:"partitionedGroupID"`
-	PartitionID        int         `json:"partitionID"`
+	// VisitTime is a unix timestamp of when the partitioning group plan was created, in order to validate if the marker
+	// is referring to the latest version of the group plan
+	PartitionedGroupCreationTime int64 `json:"partitionedGroupCreationTime"`
+	PartitionID                  int   `json:"partitionID"`
 	// VisitTime is a unix timestamp of when the partition was visited (mark updated).
 	VisitTime int64 `json:"visitTime"`
 	// Version of the file.
 	Version int `json:"version"`
 }
 
-func newPartitionVisitMarker(compactorID string, partitionedGroupID uint32, partitionID int) *partitionVisitMarker {
+func newPartitionVisitMarker(compactorID string, partitionedGroupID uint32, partitioned_group_creation_time int64, partitionID int) *partitionVisitMarker {
 	return &partitionVisitMarker{
-		CompactorID:        compactorID,
-		PartitionedGroupID: partitionedGroupID,
-		PartitionID:        partitionID,
+		CompactorID:                  compactorID,
+		PartitionedGroupID:           partitionedGroupID,
+		PartitionedGroupCreationTime: partitioned_group_creation_time,
+		PartitionID:                  partitionID,
 	}
 }
 

pkg/compactor/partitioned_group_info.go

@@ -156,7 +156,8 @@ func (p *PartitionedGroupInfo) getPartitionedGroupStatus(
 			status.PendingPartitions++
 			allPartitionCompleted = false
 			status.PendingOrFailedPartitions = append(status.PendingOrFailedPartitions, partition)
-		} else if visitMarker.VisitTime < p.CreationTime {
+		} else if visitMarker.VisitTime < p.CreationTime ||
+			(visitMarker.PartitionedGroupCreationTime > 0 && visitMarker.PartitionedGroupCreationTime < p.CreationTime) {
 			status.VisitMarkersToDelete = append(status.VisitMarkersToDelete, visitMarker)
 			allPartitionCompleted = false
 		} else if (visitMarker.GetStatus() == Pending || visitMarker.GetStatus() == InProgress) && !visitMarker.IsExpired(partitionVisitMarkerTimeout) {