X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/d3f9fad0cc83a9af47589894998a781db9c60989..5a662d84f00c0c2693c18d333bab9d0fdda7e28e:/services/datamanager/datamanager.go

diff --git a/services/datamanager/datamanager.go b/services/datamanager/datamanager.go
index ce8114ade9..d3efe62173 100644
--- a/services/datamanager/datamanager.go
+++ b/services/datamanager/datamanager.go
@@ -66,7 +66,7 @@ func singlerun() {
 	var arvLogger *logger.Logger
 	if logEventTypePrefix != "" {
 		arvLogger = logger.NewLogger(logger.LoggerParams{
-			Client: arv,
+			Client:          arv,
 			EventTypePrefix: logEventTypePrefix,
 			WriteInterval:   time.Second * time.Duration(logFrequencySeconds)})
 	}
@@ -78,7 +78,7 @@ func singlerun() {
 
 	var (
 		readCollections collection.ReadCollections
-		keepServerInfo keep.ReadServers
+		keepServerInfo  keep.ReadServers
 	)
 
 	if !summary.MaybeReadData(arvLogger, &readCollections, &keepServerInfo) {
@@ -87,8 +87,8 @@ func singlerun() {
 		go func() {
 			collectionChannel <- collection.GetCollectionsAndSummarize(
 				collection.GetCollectionsParams{
-					Client: arv,
-					Logger: arvLogger,
+					Client:    arv,
+					Logger:    arvLogger,
 					BatchSize: 50})
 		}()
 
@@ -96,36 +96,41 @@ func singlerun() {
 			keep.GetKeepServersParams{
 				Client: arv,
 				Logger: arvLogger,
-				Limit: 1000})
+				Limit:  1000})
 
 		readCollections = <-collectionChannel
 	}
 
 	summary.MaybeWriteData(arvLogger, readCollections, keepServerInfo)
 
-	replicationSummary :=
-		summary.SummarizeReplication(arvLogger, readCollections, keepServerInfo)
-
-	log.Printf("Replication Counts:" +
-		"\nBlocks In Collections: %d, " +
-		"\nBlocks In Keep: %d, " +
-		"\nMissing From Keep: %d, " +
-		"\nUnder Replicated: %d, " +
-		"\nOver Replicated: %d, " +
-		"\nReplicated Just Right: %d, " +
-		"\nNot In Any Collection: %d.",
+	buckets := summary.BucketReplication(readCollections, keepServerInfo)
+	bucketCounts := buckets.Counts()
+
+	replicationSummary := buckets.SummarizeBuckets(readCollections)
+	replicationCounts := replicationSummary.ComputeCounts()
+
+	log.Printf("Blocks In Collections: %d, "+
+		"\nBlocks In Keep: %d.",
 		len(readCollections.BlockToReplication),
-		len(keepServerInfo.BlockToServers),
-		len(replicationSummary.CollectionBlocksNotInKeep),
-		len(replicationSummary.UnderReplicatedBlocks),
-		len(replicationSummary.OverReplicatedBlocks),
-		len(replicationSummary.CorrectlyReplicatedBlocks),
-		len(replicationSummary.KeepBlocksNotInCollections))
+		len(keepServerInfo.BlockToServers))
+	log.Println(replicationCounts.PrettyPrint())
+
+	log.Printf("Blocks Histogram:")
+	for _, rlbss := range bucketCounts {
+		log.Printf("%+v: %10d",
+			rlbss.Levels,
+			rlbss.Count)
+	}
 
 	// Log that we're finished. We force the recording, since go will
-	// not wait for the timer before exiting.
+	// not wait for the write timer before exiting.
 	if arvLogger != nil {
 		arvLogger.FinalUpdate(func(p map[string]interface{}, e map[string]interface{}) {
+			summaryInfo := logger.GetOrCreateMap(p, "summary_info")
+			summaryInfo["block_replication_counts"] = bucketCounts
+			summaryInfo["replication_summary"] = replicationCounts
+			p["summary_info"] = summaryInfo
+
 			p["run_info"].(map[string]interface{})["finished_at"] = time.Now()
 		})
 	}