import (
"flag"
+ "fmt"
"git.curoverse.com/arvados.git/sdk/go/arvadosclient"
+ "git.curoverse.com/arvados.git/sdk/go/keepclient"
"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"
+ "git.curoverse.com/arvados.git/services/datamanager/summary"
"log"
"time"
)
var arvLogger *logger.Logger
if logEventTypePrefix != "" {
- arvLogger = logger.NewLogger(logger.LoggerParams{Client: arv,
+ arvLogger = logger.NewLogger(logger.LoggerParams{
+ Client: arv,
EventTypePrefix: logEventTypePrefix,
WriteInterval: time.Second * time.Duration(logFrequencySeconds)})
}
arvLogger.AddWriteHook(loggerutil.LogMemoryAlloc)
}
- collectionChannel := make(chan collection.ReadCollections)
+ var (
+ dataFetcher summary.DataFetcher
+ readCollections collection.ReadCollections
+ keepServerInfo keep.ReadServers
+ )
- go func() {
- collectionChannel <- collection.GetCollectionsAndSummarize(
- collection.GetCollectionsParams{
- Client: arv, Logger: arvLogger, BatchSize: 50})
- }()
+ if summary.ShouldReadData() {
+ dataFetcher = summary.ReadData
+ } else {
+ dataFetcher = BuildDataFetcher(arv)
+ }
+
+ dataFetcher(arvLogger, &readCollections, &keepServerInfo)
+
+ summary.MaybeWriteData(arvLogger, readCollections, keepServerInfo)
+
+ buckets := summary.BucketReplication(readCollections, keepServerInfo)
+ bucketCounts := buckets.Counts()
- keepServerInfo := keep.GetKeepServersAndSummarize(
- keep.GetKeepServersParams{Client: arv, Logger: arvLogger, Limit: 1000})
+ replicationSummary := buckets.SummarizeBuckets(readCollections)
+ replicationCounts := replicationSummary.ComputeCounts()
- readCollections := <-collectionChannel
+ log.Printf("Blocks In Collections: %d, "+
+ "\nBlocks In Keep: %d.",
+ len(readCollections.BlockToDesiredReplication),
+ len(keepServerInfo.BlockToServers))
+ log.Println(replicationCounts.PrettyPrint())
+
+ log.Printf("Blocks Histogram:")
+ for _, rlbss := range bucketCounts {
+ log.Printf("%+v: %10d",
+ rlbss.Levels,
+ rlbss.Count)
+ }
+
+ kc, err := keepclient.MakeKeepClient(&arv)
+ if err != nil {
+ loggerutil.FatalWithMessage(arvLogger,
+ fmt.Sprintf("Error setting up keep client %s", err.Error()))
+ }
- // TODO(misha): Use these together to verify replication.
- _ = readCollections
- _ = keepServerInfo
+ pullServers := summary.ComputePullServers(kc,
+ &keepServerInfo,
+ readCollections.BlockToDesiredReplication,
+ replicationSummary.UnderReplicatedBlocks)
+
+ pullLists := summary.BuildPullLists(pullServers)
+
+ summary.WritePullLists(arvLogger, pullLists)
// 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()
})
}
}
+
+// Returns a data fetcher that fetches data from remote servers.
+func BuildDataFetcher(arv arvadosclient.ArvadosClient) summary.DataFetcher {
+ return func(arvLogger *logger.Logger,
+ readCollections *collection.ReadCollections,
+ keepServerInfo *keep.ReadServers) {
+ 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
+ }
+}