X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/fb62ab318be2202b9d403e65d6dc86a9d7e72a3a..43538243995267c417983360d226d6e8eb181139:/services/datamanager/datamanager.go

diff --git a/services/datamanager/datamanager.go b/services/datamanager/datamanager.go
index 6393787e01..bd68db112a 100644
--- a/services/datamanager/datamanager.go
+++ b/services/datamanager/datamanager.go
@@ -5,12 +5,31 @@ package main
 import (
 	"flag"
 	"git.curoverse.com/arvados.git/sdk/go/arvadosclient"
+	"git.curoverse.com/arvados.git/sdk/go/logger"
 	"git.curoverse.com/arvados.git/sdk/go/util"
 	"git.curoverse.com/arvados.git/services/datamanager/collection"
-//	"git.curoverse.com/arvados.git/services/datamanager/keep"
+	"git.curoverse.com/arvados.git/services/datamanager/keep"
+	"git.curoverse.com/arvados.git/services/datamanager/loggerutil"
 	"log"
+	"time"
 )
 
+var (
+	logEventType        string
+	logFrequencySeconds int
+)
+
+func init() {
+	flag.StringVar(&logEventType,
+		"log-event-type",
+		"experimental-data-manager-report",
+		"event_type to use in our arvados log entries. Set to empty to turn off logging")
+	flag.IntVar(&logFrequencySeconds,
+		"log-frequency-seconds",
+		20,
+		"How frequently we'll write log entries in seconds.")
+}
+
 func main() {
 	flag.Parse()
 
@@ -25,38 +44,40 @@ func main() {
 		log.Fatalf("Current user is not an admin. Datamanager can only be run by admins.")
 	}
 
-	// TODO(misha): Read Collections and Keep Contents concurrently as goroutines.
-
-	readCollections := collection.GetCollections(
-		collection.GetCollectionsParams{
-			Client: arv, BatchSize: 500})
+	var arvLogger *logger.Logger
+	if logEventType != "" {
+		arvLogger = logger.NewLogger(logger.LoggerParams{Client: arv,
+			EventType:     logEventType,
+			WriteInterval: time.Second * time.Duration(logFrequencySeconds)})
+	}
 
-	//log.Printf("Read Collections: %v", readCollections)
+	loggerutil.LogRunInfo(arvLogger)
+	if arvLogger != nil {
+		arvLogger.AddWriteHook(loggerutil.LogMemoryAlloc)
+	}
 
-	UserUsage := ComputeSizeOfOwnedCollections(readCollections)
-	log.Printf("Uuid to Size used: %v", UserUsage)
+	collectionChannel := make(chan collection.ReadCollections)
 
-	// TODO(misha): Add a "readonly" flag. If we're in readonly mode,
-	// lots of behaviors can become warnings (and obviously we can't
-	// write anything).
-	// if !readCollections.ReadAllCollections {
-	// 	log.Fatalf("Did not read all collections")
-	// }
+	go func() {
+		collectionChannel <- collection.GetCollectionsAndSummarize(
+			collection.GetCollectionsParams{
+				Client: arv, Logger: arvLogger, BatchSize: 50})
+	}()
 
-	log.Printf("Read and processed %d collections",
-		len(readCollections.UuidToCollection))
+	keepServerInfo := keep.GetKeepServersAndSummarize(
+		keep.GetKeepServersParams{Client: arv, Logger: arvLogger, Limit: 1000})
 
-	// readServers := keep.GetKeepServers(
-	// 	keep.GetKeepServersParams{Client: arv, Limit: 1000})
+	readCollections := <-collectionChannel
 
-	// log.Printf("Returned %d keep disks", len(readServers.AddressToContents))
-}
+	// TODO(misha): Use these together to verify replication.
+	_ = readCollections
+	_ = keepServerInfo
 
-func ComputeSizeOfOwnedCollections(readCollections collection.ReadCollections) (
-	results map[string]int) {
-	results = make(map[string]int)
-	for _, coll := range readCollections.UuidToCollection {
-		results[coll.OwnerUuid] = results[coll.OwnerUuid] + coll.TotalSize
+	// Log that we're finished. We force the recording, since go will
+	// not wait for the timer before exiting.
+	if arvLogger != nil {
+		arvLogger.FinalUpdate(func(p map[string]interface{}, e map[string]interface{}) {
+			p["run_info"].(map[string]interface{})["time_finished"] = time.Now()
+		})
 	}
-	return
 }