X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/b17b1ad22bd3cf518eb1b3657f62768b60c7db25..d89fc97f4a7ee74edcb1a5856fec560c748b9086:/services/datamanager/keep/keep.go

diff --git a/services/datamanager/keep/keep.go b/services/datamanager/keep/keep.go
index c16512069e..651c869ef0 100644
--- a/services/datamanager/keep/keep.go
+++ b/services/datamanager/keep/keep.go
@@ -83,7 +83,7 @@ func init() {
 	flag.StringVar(&serviceType,
 		"service-type",
 		"disk",
-		"Supported keepservice type. Default is disk.")
+		"Operate only on keep_services with the specified service_type, ignoring all others.")
 }
 
 // String
@@ -118,7 +118,7 @@ func GetKeepServersAndSummarize(params GetKeepServersParams) (results ReadServer
 // GetKeepServers from api server
 func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error) {
 	sdkParams := arvadosclient.Dict{
-		"filters": [][]string{[]string{"service_type", "!=", "proxy"}},
+		"filters": [][]string{{"service_type", "!=", "proxy"}},
 	}
 	if params.Limit > 0 {
 		sdkParams["limit"] = params.Limit
@@ -131,18 +131,17 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error
 		return
 	}
 
-	// Currently, only "disk" types are supported. Stop if any other service types are found.
-	foundSupportedServieType := false
+	var keepServers []ServerAddress
 	for _, server := range sdkResponse.KeepServers {
 		if server.ServiceType == serviceType {
-			foundSupportedServieType = true
+			keepServers = append(keepServers, server)
 		} else {
-			log.Printf("Ignore unsupported service type: %v", server.ServiceType)
+			log.Printf("Skipping keep_service %q because its service_type %q does not match -service-type=%q", server, server.ServiceType, serviceType)
 		}
 	}
 
-	if !foundSupportedServieType {
-		return results, fmt.Errorf("Found no keepservices with the supported type %v", serviceType)
+	if len(keepServers) == 0 {
+		return results, fmt.Errorf("Found no keepservices with the service type %v", serviceType)
 	}
 
 	if params.Logger != nil {
@@ -151,6 +150,7 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error
 			keepInfo["num_keep_servers_available"] = sdkResponse.ItemsAvailable
 			keepInfo["num_keep_servers_received"] = len(sdkResponse.KeepServers)
 			keepInfo["keep_servers"] = sdkResponse.KeepServers
+			keepInfo["indexable_keep_servers"] = keepServers
 		})
 	}
 
@@ -160,7 +160,7 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error
 		return results, fmt.Errorf("Did not receive all available keep servers: %+v", sdkResponse)
 	}
 
-	results.KeepServerIndexToAddress = sdkResponse.KeepServers
+	results.KeepServerIndexToAddress = keepServers
 	results.KeepServerAddressToIndex = make(map[ServerAddress]int)
 	for i, address := range results.KeepServerIndexToAddress {
 		results.KeepServerAddressToIndex[address] = i
@@ -170,7 +170,7 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error
 
 	// Send off all the index requests concurrently
 	responseChan := make(chan ServerResponse)
-	for _, keepServer := range sdkResponse.KeepServers {
+	for _, keepServer := range results.KeepServerIndexToAddress {
 		// The above keepsServer variable is reused for each iteration, so
 		// it would be shared across all goroutines. This would result in
 		// us querying one server n times instead of n different servers
@@ -188,7 +188,7 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error
 	results.BlockToServers = make(map[blockdigest.DigestWithSize][]BlockServerInfo)
 
 	// Read all the responses
-	for i := range sdkResponse.KeepServers {
+	for i := range results.KeepServerIndexToAddress {
 		_ = i // Here to prevent go from complaining.
 		response := <-responseChan
 
@@ -430,13 +430,23 @@ func parseBlockInfoFromIndexLine(indexLine string) (blockInfo BlockInfo, err err
 		return
 	}
 
-	blockInfo.Mtime, err = strconv.ParseInt(tokens[1], 10, 64)
+	var ns int64
+	ns, err = strconv.ParseInt(tokens[1], 10, 64)
 	if err != nil {
 		return
 	}
-	blockInfo.Digest =
-		blockdigest.DigestWithSize{Digest: locator.Digest,
-			Size: uint32(locator.Size)}
+	if ns < 1e12 {
+		// An old version of keepstore is giving us timestamps
+		// in seconds instead of nanoseconds. (This threshold
+		// correctly handles all times between 1970-01-02 and
+		// 33658-09-27.)
+		ns = ns * 1e9
+	}
+	blockInfo.Mtime = ns
+	blockInfo.Digest = blockdigest.DigestWithSize{
+		Digest: locator.Digest,
+		Size:   uint32(locator.Size),
+	}
 	return
 }
 
@@ -466,13 +476,29 @@ type TrashRequest struct {
 type TrashList []TrashRequest
 
 // SendTrashLists to trash queue
-func SendTrashLists(kc *keepclient.KeepClient, spl map[string]TrashList) (errs []error) {
+func SendTrashLists(arvLogger *logger.Logger, kc *keepclient.KeepClient, spl map[string]TrashList, dryRun bool) (errs []error) {
 	count := 0
 	barrier := make(chan error)
 
 	client := kc.Client
 
 	for url, v := range spl {
+		if arvLogger != nil {
+			// We need a local variable because Update doesn't call our mutator func until later,
+			// when our list variable might have been reused by the next loop iteration.
+			url := url
+			trashLen := len(v)
+			arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
+				trashListInfo := logger.GetOrCreateMap(p, "trash_list_len")
+				trashListInfo[url] = trashLen
+			})
+		}
+
+		if dryRun {
+			log.Printf("dry run, not sending trash list to service %s with %d blocks", url, len(v))
+			continue
+		}
+
 		count++
 		log.Printf("Sending trash list to %v", url)
 
@@ -512,7 +538,6 @@ func SendTrashLists(kc *keepclient.KeepClient, spl map[string]TrashList) (errs [
 				barrier <- nil
 			}
 		})(url, v)
-
 	}
 
 	for i := 0; i < count; i++ {