Persist in-memory replication stats to disk (#15594)

to avoid relying on scanner-calculated replication metrics.
This will improve the accuracy of the replication stats reported.

This PR also adds on to #15556 by handing replication
traffic that could not be queued by available workers to the 
MRF queue so that entries in `PENDING` status are healed faster.

cmd/bucket-replication-stats.go

@@ -19,6 +19,7 @@ package cmd
 
 import (
 	"context"
+	"math"
 	"sync"
 	"time"
 
@@ -36,10 +37,12 @@ func (b *BucketReplicationStats) hasReplicationUsage() bool {
 
 // ReplicationStats holds the global in-memory replication stats
 type ReplicationStats struct {
-	Cache      map[string]*BucketReplicationStats
-	UsageCache map[string]*BucketReplicationStats
-	sync.RWMutex
-	ulock sync.RWMutex
+	Cache           map[string]*BucketReplicationStats
+	UsageCache      map[string]*BucketReplicationStats
+	mostRecentStats BucketStatsMap
+	sync.RWMutex                 // mutex for Cache
+	ulock           sync.RWMutex // mutex for UsageCache
+	dlock           sync.RWMutex // mutex for mostRecentStats
 }
 
 // Delete deletes in-memory replication statistics for a bucket.
@@ -92,14 +95,23 @@ func (r *ReplicationStats) Update(bucket string, arn string, n int64, duration t
 		bs.Stats[arn] = b
 	}
 	switch status {
+	case replication.Pending:
+		if opType.IsDataReplication() && prevStatus != status {
+			b.PendingSize += n
+			b.PendingCount++
+		}
 	case replication.Completed:
 		switch prevStatus { // adjust counters based on previous state
+		case replication.Pending:
+			b.PendingCount--
 		case replication.Failed:
 			b.FailedCount--
 		}
-		if opType == replication.ObjectReplicationType {
+		if opType.IsDataReplication() {
 			b.ReplicatedSize += n
 			switch prevStatus {
+			case replication.Pending:
+				b.PendingSize -= n
 			case replication.Failed:
 				b.FailedSize -= n
 			}
@@ -108,10 +120,12 @@ func (r *ReplicationStats) Update(bucket string, arn string, n int64, duration t
 			}
 		}
 	case replication.Failed:
-		if opType == replication.ObjectReplicationType {
+		if opType.IsDataReplication() {
 			if prevStatus == replication.Pending {
 				b.FailedSize += n
 				b.FailedCount++
+				b.PendingSize -= n
+				b.PendingCount--
 			}
 		}
 	case replication.Replica:
@@ -176,9 +190,20 @@ func NewReplicationStats(ctx context.Context, objectAPI ObjectLayer) *Replicatio
 	}
 }
 
-// load replication metrics at cluster start from initial data usage
+// load replication metrics at cluster start from latest replication stats saved in .minio.sys/buckets/replication/node-name.stats
+// fallback to replication stats in data usage to be backward compatible
 func (r *ReplicationStats) loadInitialReplicationMetrics(ctx context.Context) {
-	rTimer := time.NewTimer(time.Minute)
+	m := make(map[string]*BucketReplicationStats)
+	if stats, err := globalReplicationPool.loadStatsFromDisk(); err == nil {
+		for b, st := range stats {
+			m[b] = &st
+		}
+		r.ulock.Lock()
+		r.UsageCache = m
+		r.ulock.Unlock()
+		return
+	}
+	rTimer := time.NewTimer(time.Second * 5)
 	defer rTimer.Stop()
 	var (
 		dui DataUsageInfo
@@ -192,15 +217,12 @@ outer:
 		case <-rTimer.C:
 			dui, err = loadDataUsageFromBackend(GlobalContext, newObjectLayerFn())
 			// If LastUpdate is set, data usage is available.
-			if err == nil && !dui.LastUpdate.IsZero() {
+			if err == nil {
 				break outer
 			}
-
-			rTimer.Reset(time.Minute)
+			rTimer.Reset(time.Second * 5)
 		}
 	}
-
-	m := make(map[string]*BucketReplicationStats)
 	for bucket, usage := range dui.BucketsUsage {
 		b := &BucketReplicationStats{
 			Stats: make(map[string]*BucketReplicationStat, len(usage.ReplicationInfo)),
@@ -219,6 +241,123 @@ outer:
 		}
 	}
 	r.ulock.Lock()
-	defer r.ulock.Unlock()
 	r.UsageCache = m
+	r.ulock.Unlock()
+}
+
+func (r *ReplicationStats) getAllCachedLatest() BucketStatsMap {
+	r.dlock.RLock()
+	defer r.dlock.RUnlock()
+	return r.mostRecentStats
+}
+
+func (r *ReplicationStats) getAllLatest(bucketsUsage map[string]BucketUsageInfo) (bucketsReplicationStats map[string]BucketReplicationStats) {
+	peerBucketStatsList := globalNotificationSys.GetClusterAllBucketStats(GlobalContext)
+	bucketsReplicationStats = make(map[string]BucketReplicationStats, len(bucketsUsage))
+
+	for bucket, u := range bucketsUsage {
+		bucketStats := make([]BucketStats, len(peerBucketStatsList))
+		for i, peerBucketStats := range peerBucketStatsList {
+			bucketStat, ok := peerBucketStats.Stats[bucket]
+			if !ok {
+				continue
+			}
+			bucketStats[i] = bucketStat
+		}
+		bucketsReplicationStats[bucket] = r.calculateBucketReplicationStats(bucket, u, bucketStats)
+	}
+	return bucketsReplicationStats
+}
+
+func (r *ReplicationStats) calculateBucketReplicationStats(bucket string, u BucketUsageInfo, bucketStats []BucketStats) (s BucketReplicationStats) {
+	// accumulate cluster bucket stats
+	stats := make(map[string]*BucketReplicationStat)
+	var totReplicaSize int64
+	for _, bucketStat := range bucketStats {
+		totReplicaSize += bucketStat.ReplicationStats.ReplicaSize
+		for arn, stat := range bucketStat.ReplicationStats.Stats {
+			oldst := stats[arn]
+			if oldst == nil {
+				oldst = &BucketReplicationStat{}
+			}
+			stats[arn] = &BucketReplicationStat{
+				FailedCount:    stat.FailedCount + oldst.FailedCount,
+				FailedSize:     stat.FailedSize + oldst.FailedSize,
+				ReplicatedSize: stat.ReplicatedSize + oldst.ReplicatedSize,
+				Latency:        stat.Latency.merge(oldst.Latency),
+				PendingCount:   stat.PendingCount + oldst.PendingCount,
+				PendingSize:    stat.PendingSize + oldst.PendingSize,
+			}
+		}
+	}
+
+	// add initial usage stat to cluster stats
+	usageStat := globalReplicationStats.GetInitialUsage(bucket)
+
+	totReplicaSize += usageStat.ReplicaSize
+	for arn, stat := range usageStat.Stats {
+		st, ok := stats[arn]
+		if !ok {
+			st = &BucketReplicationStat{}
+			stats[arn] = st
+		}
+		st.ReplicatedSize += stat.ReplicatedSize
+		st.FailedSize += stat.FailedSize
+		st.FailedCount += stat.FailedCount
+		st.PendingSize += stat.PendingSize
+		st.PendingCount += stat.PendingCount
+	}
+
+	s = BucketReplicationStats{
+		Stats: make(map[string]*BucketReplicationStat, len(stats)),
+	}
+	var latestTotReplicatedSize int64
+	for _, st := range u.ReplicationInfo {
+		latestTotReplicatedSize += int64(st.ReplicatedSize)
+	}
+
+	// normalize computed real time stats with latest usage stat
+	for arn, tgtstat := range stats {
+		st := BucketReplicationStat{}
+		bu, ok := u.ReplicationInfo[arn]
+		if !ok {
+			bu = BucketTargetUsageInfo{}
+		}
+		// use in memory replication stats if it is ahead of usage info.
+		st.ReplicatedSize = int64(bu.ReplicatedSize)
+		if tgtstat.ReplicatedSize >= int64(bu.ReplicatedSize) {
+			st.ReplicatedSize = tgtstat.ReplicatedSize
+		}
+		s.ReplicatedSize += st.ReplicatedSize
+		// Reset FailedSize and FailedCount to 0 for negative overflows which can
+		// happen since data usage picture can lag behind actual usage state at the time of cluster start
+		st.FailedSize = int64(math.Max(float64(tgtstat.FailedSize), 0))
+		st.FailedCount = int64(math.Max(float64(tgtstat.FailedCount), 0))
+		st.PendingSize = int64(math.Max(float64(tgtstat.PendingSize), 0))
+		st.PendingCount = int64(math.Max(float64(tgtstat.PendingCount), 0))
+		st.Latency = tgtstat.Latency
+
+		s.Stats[arn] = &st
+		s.FailedSize += st.FailedSize
+		s.FailedCount += st.FailedCount
+		s.PendingCount += st.PendingCount
+		s.PendingSize += st.PendingSize
+	}
+	// normalize overall stats
+	s.ReplicaSize = int64(math.Max(float64(totReplicaSize), float64(u.ReplicaSize)))
+	s.ReplicatedSize = int64(math.Max(float64(s.ReplicatedSize), float64(latestTotReplicatedSize)))
+	r.dlock.Lock()
+	if len(r.mostRecentStats.Stats) == 0 {
+		r.mostRecentStats = BucketStatsMap{Stats: make(map[string]BucketStats, 1), Timestamp: UTCNow()}
+	}
+	r.mostRecentStats.Stats[bucket] = BucketStats{ReplicationStats: s}
+	r.mostRecentStats.Timestamp = UTCNow()
+	r.dlock.Unlock()
+	return s
+}
+
+// get the most current of in-memory replication stats  and data usage info from crawler.
+func (r *ReplicationStats) getLatestReplicationStats(bucket string, u BucketUsageInfo) (s BucketReplicationStats) {
+	bucketStats := globalNotificationSys.GetClusterBucketStats(GlobalContext, bucket)
+	return r.calculateBucketReplicationStats(bucket, u, bucketStats)
 }