Merge branch '3198-writable-fuse' into 6219-fuse-performance-testing

[arvados.git] / services / datamanager / datamanager.go

diff --git a/services/datamanager/datamanager.go b/services/datamanager/datamanager.go
index 6bd9ee8afa8fc0de7df8bc17e93be3c3caacb2dd..a8e506eacb1d0f5c533d005a7a3b321b9c6b76d3 100644 (file)
--- a/services/datamanager/datamanager.go
+++ b/services/datamanager/datamanager.go
@@ -9,31 +9,48 @@ import (
        "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/loggerutil"
        "log"
-       "os"
-       "runtime"
        "time"
 )
 
 var (
-       logEventType string
+       logEventTypePrefix  string
        logFrequencySeconds int
+       minutesBetweenRuns  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, 
+       flag.StringVar(&logEventTypePrefix,
+               "log-event-type-prefix",
+               "experimental-data-manager",
+               "Prefix to use in the event_type of 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.")
+       flag.IntVar(&minutesBetweenRuns,
+               "minutes-between-runs",
+               0,
+               "How many minutes we wait betwen data manager runs. 0 means run once and exit.")
 }
 
 func main() {
        flag.Parse()
+       if minutesBetweenRuns == 0 {
+               singlerun()
+       } else {
+               waitTime := time.Minute * time.Duration(minutesBetweenRuns)
+               for {
+                       log.Println("Beginning Run")
+                       singlerun()
+                       log.Printf("Sleeping for %d minutes", minutesBetweenRuns)
+                       time.Sleep(waitTime)
+               }
+       }
+}
 
+func singlerun() {
        arv, err := arvadosclient.MakeArvadosClient()
        if err != nil {
                log.Fatalf("Error setting up arvados client %s", err.Error())
@@ -46,69 +63,39 @@ func main() {
        }
 
        var arvLogger *logger.Logger
-       if logEventType != "" {
+       if logEventTypePrefix != "" {
                arvLogger = logger.NewLogger(logger.LoggerParams{Client: arv,
-                       EventType: logEventType,
-                       MinimumWriteInterval: time.Second * time.Duration(logFrequencySeconds)})
+                       EventTypePrefix: logEventTypePrefix,
+                       WriteInterval:   time.Second * time.Duration(logFrequencySeconds)})
        }
 
+       loggerutil.LogRunInfo(arvLogger)
        if arvLogger != nil {
-               properties, _ := arvLogger.Edit()
-               runInfo := make(map[string]interface{})
-               runInfo["start_time"] = time.Now()
-               runInfo["args"] = os.Args
-               hostname, err := os.Hostname()
-               if err != nil {
-                       runInfo["hostname_error"] = err.Error()
-               } else {
-                       runInfo["hostname"] = hostname
-               }
-               runInfo["pid"] = os.Getpid()
-               properties["run_info"] = runInfo
-
-               arvLogger.AddWriteHook(LogMemoryAlloc)
-
-               arvLogger.Record()
+               arvLogger.AddWriteHook(loggerutil.LogMemoryAlloc)
        }
 
        collectionChannel := make(chan collection.ReadCollections)
 
-       go func() { collectionChannel <- collection.GetCollectionsAndSummarize(
-               collection.GetCollectionsParams{
-                       Client: arv, Logger: arvLogger, BatchSize: 50}) }()
+       go func() {
+               collectionChannel <- collection.GetCollectionsAndSummarize(
+                       collection.GetCollectionsParams{
+                               Client: arv, Logger: arvLogger, BatchSize: 50})
+       }()
 
-       RunKeep(keep.GetKeepServersParams{Client: arv, Limit: 1000})
+       keepServerInfo := keep.GetKeepServersAndSummarize(
+               keep.GetKeepServersParams{Client: arv, Logger: arvLogger, Limit: 1000})
 
        readCollections := <-collectionChannel
-       _ = readCollections  // Make compiler happy.
 
-       // Log that we're finished
-       if arvLogger != nil {
-               properties,_ := arvLogger.Edit()
-               properties["run_info"].(map[string]interface{})["end_time"] = time.Now()
-               // Force the recording, since go will not wait for the timer before exiting.
-               arvLogger.ForceRecord()
-       }
-}
-
-func RunKeep(params keep.GetKeepServersParams) {
-       readServers := keep.GetKeepServers(params)
+       // TODO(misha): Use these together to verify replication.
+       _ = readCollections
+       _ = keepServerInfo
 
-       log.Printf("Returned %d keep disks", len(readServers.ServerToContents))
-
-       blockReplicationCounts := make(map[int]int)
-       for _, infos := range readServers.BlockToServers {
-               replication := len(infos)
-               blockReplicationCounts[replication] += 1
+       // 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{})["finished_at"] = time.Now()
+               })
        }
-
-       log.Printf("Replication level distribution: %v", blockReplicationCounts)
-}
-
-func LogMemoryAlloc(properties map[string]interface{}, entry map[string]interface{}) {
-       _ = entry  // keep the compiler from complaining
-       runInfo := properties["run_info"].(map[string]interface{})
-       var memStats runtime.MemStats
-       runtime.ReadMemStats(&memStats)
-       runInfo["alloc_bytes_in_use"] = memStats.Alloc
 }