1 /* Keep Datamanager. Responsible for checking on and reporting on Keep Storage */
8 "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
9 "git.curoverse.com/arvados.git/sdk/go/keepclient"
10 "git.curoverse.com/arvados.git/sdk/go/logger"
11 "git.curoverse.com/arvados.git/sdk/go/util"
12 "git.curoverse.com/arvados.git/services/datamanager/collection"
13 "git.curoverse.com/arvados.git/services/datamanager/keep"
14 "git.curoverse.com/arvados.git/services/datamanager/loggerutil"
15 "git.curoverse.com/arvados.git/services/datamanager/summary"
21 logEventTypePrefix string
22 logFrequencySeconds int
23 minutesBetweenRuns int
27 flag.StringVar(&logEventTypePrefix,
28 "log-event-type-prefix",
29 "experimental-data-manager",
30 "Prefix to use in the event_type of our arvados log entries. Set to empty to turn off logging")
31 flag.IntVar(&logFrequencySeconds,
32 "log-frequency-seconds",
34 "How frequently we'll write log entries in seconds.")
35 flag.IntVar(&minutesBetweenRuns,
36 "minutes-between-runs",
38 "How many minutes we wait betwen data manager runs. 0 means run once and exit.")
43 if minutesBetweenRuns == 0 {
46 waitTime := time.Minute * time.Duration(minutesBetweenRuns)
48 log.Println("Beginning Run")
50 log.Printf("Sleeping for %d minutes", minutesBetweenRuns)
57 arv, err := arvadosclient.MakeArvadosClient()
59 log.Fatalf("Error setting up arvados client %s", err.Error())
62 if is_admin, err := util.UserIsAdmin(arv); err != nil {
63 log.Fatalf("Error querying current arvados user %s", err.Error())
65 log.Fatalf("Current user is not an admin. Datamanager can only be run by admins.")
68 var arvLogger *logger.Logger
69 if logEventTypePrefix != "" {
70 arvLogger = logger.NewLogger(logger.LoggerParams{
72 EventTypePrefix: logEventTypePrefix,
73 WriteInterval: time.Second * time.Duration(logFrequencySeconds)})
76 loggerutil.LogRunInfo(arvLogger)
78 arvLogger.AddWriteHook(loggerutil.LogMemoryAlloc)
82 dataFetcher summary.DataFetcher
83 readCollections collection.ReadCollections
84 keepServerInfo keep.ReadServers
87 if summary.ShouldReadData() {
88 dataFetcher = summary.ReadData
90 dataFetcher = BuildDataFetcher(arv)
93 dataFetcher(arvLogger, &readCollections, &keepServerInfo)
95 summary.MaybeWriteData(arvLogger, readCollections, keepServerInfo)
97 buckets := summary.BucketReplication(readCollections, keepServerInfo)
98 bucketCounts := buckets.Counts()
100 replicationSummary := buckets.SummarizeBuckets(readCollections)
101 replicationCounts := replicationSummary.ComputeCounts()
103 log.Printf("Blocks In Collections: %d, "+
104 "\nBlocks In Keep: %d.",
105 len(readCollections.BlockToDesiredReplication),
106 len(keepServerInfo.BlockToServers))
107 log.Println(replicationCounts.PrettyPrint())
109 log.Printf("Blocks Histogram:")
110 for _, rlbss := range bucketCounts {
111 log.Printf("%+v: %10d",
116 kc, err := keepclient.MakeKeepClient(&arv)
118 loggerutil.FatalWithMessage(arvLogger,
119 fmt.Sprintf("Error setting up keep client %s", err.Error()))
122 pullServers := summary.ComputePullServers(kc,
124 readCollections.BlockToDesiredReplication,
125 replicationSummary.UnderReplicatedBlocks)
127 pullLists := summary.BuildPullLists(pullServers)
128 trashLists := summary.BuildTrashLists(kc,
130 replicationSummary.KeepBlocksNotInCollections)
132 summary.WritePullLists(arvLogger, pullLists)
133 keep.SendTrashLists(arvLogger, kc, trashLists)
135 // Log that we're finished. We force the recording, since go will
136 // not wait for the write timer before exiting.
137 if arvLogger != nil {
138 arvLogger.FinalUpdate(func(p map[string]interface{}, e map[string]interface{}) {
139 summaryInfo := logger.GetOrCreateMap(p, "summary_info")
140 summaryInfo["block_replication_counts"] = bucketCounts
141 summaryInfo["replication_summary"] = replicationCounts
142 p["summary_info"] = summaryInfo
144 p["run_info"].(map[string]interface{})["finished_at"] = time.Now()
149 // Returns a data fetcher that fetches data from remote servers.
150 func BuildDataFetcher(arv arvadosclient.ArvadosClient) summary.DataFetcher {
151 return func(arvLogger *logger.Logger,
152 readCollections *collection.ReadCollections,
153 keepServerInfo *keep.ReadServers) {
154 collectionChannel := make(chan collection.ReadCollections)
157 collectionChannel <- collection.GetCollectionsAndSummarize(
158 collection.GetCollectionsParams{
164 *keepServerInfo = keep.GetKeepServersAndSummarize(
165 keep.GetKeepServersParams{
170 *readCollections = <-collectionChannel