Merge branch '10460-cwl-directory-literal' closes #10460
[arvados.git] / services / datamanager / keep / keep.go
index c49a0211a8ee842c2a51d5bb3ceb4f38cee4d9ea..39d2d5bd5d7d7b5af23675baa295e5a3eb38f2e7 100644 (file)
@@ -6,6 +6,7 @@ import (
        "bufio"
        "encoding/json"
        "errors"
+       "flag"
        "fmt"
        "git.curoverse.com/arvados.git/sdk/go/arvadosclient"
        "git.curoverse.com/arvados.git/sdk/go/blockdigest"
@@ -65,7 +66,7 @@ type ReadServers struct {
 
 // GetKeepServersParams struct
 type GetKeepServersParams struct {
-       Client arvadosclient.ArvadosClient
+       Client *arvadosclient.ArvadosClient
        Logger *logger.Logger
        Limit  int
 }
@@ -76,6 +77,15 @@ type ServiceList struct {
        KeepServers    []ServerAddress `json:"items"`
 }
 
+var serviceType string
+
+func init() {
+       flag.StringVar(&serviceType,
+               "service-type",
+               "disk",
+               "Operate only on keep_services with the specified service_type, ignoring all others.")
+}
+
 // String
 // TODO(misha): Change this to include the UUID as well.
 func (s ServerAddress) String() string {
@@ -108,7 +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
@@ -121,19 +131,26 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error
                return
        }
 
-       // Currently, only "disk" types are supported. Stop if any other service types are found.
+       var keepServers []ServerAddress
        for _, server := range sdkResponse.KeepServers {
-               if server.ServiceType != "disk" {
-                       return results, fmt.Errorf("Unsupported service type %q found for: %v", server.ServiceType, server)
+               if server.ServiceType == serviceType {
+                       keepServers = append(keepServers, server)
+               } else {
+                       log.Printf("Skipping keep_service %q because its service_type %q does not match -service-type=%q", server, server.ServiceType, serviceType)
                }
        }
 
+       if len(keepServers) == 0 {
+               return results, fmt.Errorf("Found no keepservices with the service type %v", serviceType)
+       }
+
        if params.Logger != nil {
                params.Logger.Update(func(p map[string]interface{}, e map[string]interface{}) {
                        keepInfo := logger.GetOrCreateMap(p, "keep_info")
                        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
                })
        }
 
@@ -143,7 +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
@@ -153,7 +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
@@ -171,7 +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
 
@@ -198,7 +215,7 @@ func GetKeepServers(params GetKeepServersParams) (results ReadServers, err error
 // GetServerContents of the keep server
 func GetServerContents(arvLogger *logger.Logger,
        keepServer ServerAddress,
-       arv arvadosclient.ArvadosClient) (response ServerResponse) {
+       arv *arvadosclient.ArvadosClient) (response ServerResponse) {
 
        err := GetServerStatus(arvLogger, keepServer, arv)
        if err != nil {
@@ -230,7 +247,7 @@ func GetServerContents(arvLogger *logger.Logger,
 // GetServerStatus get keep server status by invoking /status.json
 func GetServerStatus(arvLogger *logger.Logger,
        keepServer ServerAddress,
-       arv arvadosclient.ArvadosClient) error {
+       arv *arvadosclient.ArvadosClient) error {
        url := fmt.Sprintf("http://%s:%d/status.json",
                keepServer.Host,
                keepServer.Port)
@@ -281,7 +298,7 @@ func GetServerStatus(arvLogger *logger.Logger,
 // CreateIndexRequest to the keep server
 func CreateIndexRequest(arvLogger *logger.Logger,
        keepServer ServerAddress,
-       arv arvadosclient.ArvadosClient) (req *http.Request, err error) {
+       arv *arvadosclient.ArvadosClient) (req *http.Request, err error) {
        url := fmt.Sprintf("http://%s:%d/index", keepServer.Host, keepServer.Port)
        log.Println("About to fetch keep server contents from " + url)
 
@@ -413,13 +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
 }
 
@@ -456,57 +483,61 @@ func SendTrashLists(arvLogger *logger.Logger, kc *keepclient.KeepClient, spl map
        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 {
-                       if arvLogger != nil {
-                               for url, v := range spl {
-                                       arvLogger.Update(func(p map[string]interface{}, e map[string]interface{}) {
-                                               trashListInfo := logger.GetOrCreateMap(p, "trash_list")
-                                               trashListInfo["started_at"] = time.Now()
-                                               trashListInfo[url] = v
-                                       })
-                               }
-                       }
-               } else {
-                       count++
-                       log.Printf("Sending trash list to %v", url)
-
-                       go (func(url string, v TrashList) {
-                               pipeReader, pipeWriter := io.Pipe()
-                               go (func() {
-                                       enc := json.NewEncoder(pipeWriter)
-                                       enc.Encode(v)
-                                       pipeWriter.Close()
-                               })()
-
-                               req, err := http.NewRequest("PUT", fmt.Sprintf("%s/trash", url), pipeReader)
-                               if err != nil {
-                                       log.Printf("Error creating trash list request for %v error: %v", url, err.Error())
-                                       barrier <- err
-                                       return
-                               }
-
-                               req.Header.Add("Authorization", "OAuth2 "+kc.Arvados.ApiToken)
-
-                               // Make the request
-                               var resp *http.Response
-                               if resp, err = client.Do(req); err != nil {
-                                       log.Printf("Error sending trash list to %v error: %v", url, err.Error())
-                                       barrier <- err
-                                       return
-                               }
-
-                               log.Printf("Sent trash list to %v: response was HTTP %v", url, resp.Status)
-
-                               io.Copy(ioutil.Discard, resp.Body)
-                               resp.Body.Close()
-
-                               if resp.StatusCode != 200 {
-                                       barrier <- errors.New(fmt.Sprintf("Got HTTP code %v", resp.StatusCode))
-                               } else {
-                                       barrier <- nil
-                               }
-                       })(url, v)
+                       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)
+
+               go (func(url string, v TrashList) {
+                       pipeReader, pipeWriter := io.Pipe()
+                       go (func() {
+                               enc := json.NewEncoder(pipeWriter)
+                               enc.Encode(v)
+                               pipeWriter.Close()
+                       })()
+
+                       req, err := http.NewRequest("PUT", fmt.Sprintf("%s/trash", url), pipeReader)
+                       if err != nil {
+                               log.Printf("Error creating trash list request for %v error: %v", url, err.Error())
+                               barrier <- err
+                               return
+                       }
+
+                       req.Header.Add("Authorization", "OAuth2 "+kc.Arvados.ApiToken)
+
+                       // Make the request
+                       var resp *http.Response
+                       if resp, err = client.Do(req); err != nil {
+                               log.Printf("Error sending trash list to %v error: %v", url, err.Error())
+                               barrier <- err
+                               return
+                       }
+
+                       log.Printf("Sent trash list to %v: response was HTTP %v", url, resp.Status)
+
+                       io.Copy(ioutil.Discard, resp.Body)
+                       resp.Body.Close()
+
+                       if resp.StatusCode != 200 {
+                               barrier <- errors.New(fmt.Sprintf("Got HTTP code %v", resp.StatusCode))
+                       } else {
+                               barrier <- nil
+                       }
+               })(url, v)
        }
 
        for i := 0; i < count; i++ {