Merge branch 'master' into 6156-hostnames-in-nodes
[arvados.git] / services / datamanager / datamanager.go
index e00f103da6d941fbb9cfa9e0cd0f705d55588b9c..a8e506eacb1d0f5c533d005a7a3b321b9c6b76d3 100644 (file)
@@ -3,40 +3,99 @@
 package main
 
 import (
-       //"git.curoverse.com/arvados.git/sdk/go/keepclient"
-       "fmt"
+       "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/loggerutil"
        "log"
+       "time"
 )
 
-// Helper type so we don't have to write out 'map[string]interface{}' every time.
-type Dict map[string]interface{}
+var (
+       logEventTypePrefix  string
+       logFrequencySeconds int
+       minutesBetweenRuns  int
+)
 
-func UserIsAdmin(arv arvadosclient.ArvadosClient) (is_admin bool, err error) {
-       type user struct {
-               IsAdmin bool `json:"is_admin"`
-       }
-       var u user
-       err = arv.Call("GET", "users", "", "current", nil, &u)
-       return u.IsAdmin, err
+func init() {
+       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() {
-       fmt.Println("Hello, world\n")
+       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())
        }
 
-       if is_admin, err := UserIsAdmin(arv); err != nil {
+       if is_admin, err := util.UserIsAdmin(arv); err != nil {
                log.Fatalf("Error querying current arvados user %s", err.Error())
        } else if !is_admin {
                log.Fatalf("Current user is not an admin. Datamanager can only be run by admins.")
        }
 
-       readCollections := collection.GetCollections(arv)
+       var arvLogger *logger.Logger
+       if logEventTypePrefix != "" {
+               arvLogger = logger.NewLogger(logger.LoggerParams{Client: arv,
+                       EventTypePrefix: logEventTypePrefix,
+                       WriteInterval:   time.Second * time.Duration(logFrequencySeconds)})
+       }
+
+       loggerutil.LogRunInfo(arvLogger)
+       if arvLogger != nil {
+               arvLogger.AddWriteHook(loggerutil.LogMemoryAlloc)
+       }
 
-       log.Printf("Read Collections: %v", readCollections)
+       collectionChannel := make(chan collection.ReadCollections)
+
+       go func() {
+               collectionChannel <- collection.GetCollectionsAndSummarize(
+                       collection.GetCollectionsParams{
+                               Client: arv, Logger: arvLogger, BatchSize: 50})
+       }()
+
+       keepServerInfo := keep.GetKeepServersAndSummarize(
+               keep.GetKeepServersParams{Client: arv, Logger: arvLogger, Limit: 1000})
+
+       readCollections := <-collectionChannel
+
+       // TODO(misha): Use these together to verify replication.
+       _ = readCollections
+       _ = keepServerInfo
+
+       // 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()
+               })
+       }
 }