diff --git a/cmd/dotmesh-server/pkg/main/rpc.go b/cmd/dotmesh-server/pkg/main/rpc.go
index 6bc42e311..6994f921d 100644
--- a/cmd/dotmesh-server/pkg/main/rpc.go
+++ b/cmd/dotmesh-server/pkg/main/rpc.go
@@ -1,7 +1,6 @@
 package main
 
 import (
-	"github.com/aws/aws-sdk-go"
 	"encoding/json"
 	"fmt"
 	"log"
@@ -11,6 +10,10 @@ import (
 	"strings"
 	"time"
 
+	"github.com/aws/aws-sdk-go/aws"
+	"github.com/aws/aws-sdk-go/aws/credentials"
+	"github.com/aws/aws-sdk-go/aws/session"
+	"github.com/aws/aws-sdk-go/service/s3"
 	"github.com/coreos/etcd/client"
 	"golang.org/x/net/context"
 )
@@ -1042,69 +1045,11 @@ func (d *DotmeshRPC) GetTransfer(
 	return nil
 }
 
-// Register a transfer from an initiator (the cluster where the user initially
-// connected) to a peer (the cluster which will be the target of a push/pull).
-func (d *DotmeshRPC) RegisterTransfer(
-	r *http.Request,
-	args *TransferPollResult,
-	result *bool,
-) error {
-	log.Printf("[RegisterTransfer] called with args: %+v", args)
-	serialized, err := json.Marshal(args)
-	if err != nil {
-		return err
-	}
-	kapi, err := getEtcdKeysApi()
-	if err != nil {
-		return err
-	}
-
-	_, err = kapi.Set(
-		context.Background(),
-		fmt.Sprintf(
-			"%s/filesystems/transfers/%s", ETCD_PREFIX, args.TransferRequestId,
-		),
-		string(serialized),
-		nil,
-	)
-	if err != nil {
-		return err
-	}
-	// XXX A transfer should be able to span multiple filesystemIds, really. So
-	// tying a transfer to a filesystem id is probably wrong. except, the thing
-	// being updated is a specific branch (filesystem id), it's ok if it drags
-	// dependent snapshots along with it.
-	responseChan, err := d.state.globalFsRequest(args.FilesystemId, &Event{
-		Name: "peer-transfer",
-		Args: &EventArgs{
-			"Transfer": args,
-		},
-	})
-	if err != nil {
-		return err
-	}
-
-	// Block until the fsmachine is ready to transfer
-	log.Printf("[RegisterTransfer:%s] waiting for ack from the fsmachine...", args.FilesystemId)
-	e := <-responseChan
-	log.Printf("[RegisterTransfer:%s] received ack from the fsmachine: %+v", args.FilesystemId, e)
-
-	if e.Name != "awaiting-transfer" {
-		// Something went wrong!
-		return fmt.Errorf("Error requesting peer transfer: %+v", e)
-	} else {
-		return nil
-	}
-
-	return nil
-}
-
 func (d *DotmeshRPC) S3Transfer(
 	r *http.Request,
 	args *S3TransferRequest,
 	result *string,
-) error
-{
+) error {
 	switch args.Direction {
 	case "push":
 		// TODO do we need this check, should it be different? Suspect s3 just requires there's no `.` etc but that's something the user will need to do on AWS anyway, unless we're allowing create-if-not-exists for pushes
@@ -1161,10 +1106,9 @@ func (d *DotmeshRPC) S3Transfer(
 			return err
 		}
 		filesystemId = remoteFilesystemId
-	} 
+	}
 	// TODO: look at the logic for filesystem combinations, there was a bunch of logic for pulling/pushing fsystems that were in sync or had dirty changes. Not sure we need or can use those
 
-	
 	// Now run globalFsRequest, returning the request id, to make the master of
 	// a (possibly nonexisting) filesystem start pulling or pushing it, and
 	// make it update status as it goes in a new pollable "transfers" object in
@@ -1191,6 +1135,64 @@ func (d *DotmeshRPC) S3Transfer(
 	*result = requestId
 	return nil
 }
+
+// Register a transfer from an initiator (the cluster where the user initially
+// connected) to a peer (the cluster which will be the target of a push/pull).
+func (d *DotmeshRPC) RegisterTransfer(
+	r *http.Request,
+	args *TransferPollResult,
+	result *bool,
+) error {
+	log.Printf("[RegisterTransfer] called with args: %+v", args)
+	serialized, err := json.Marshal(args)
+	if err != nil {
+		return err
+	}
+	kapi, err := getEtcdKeysApi()
+	if err != nil {
+		return err
+	}
+
+	_, err = kapi.Set(
+		context.Background(),
+		fmt.Sprintf(
+			"%s/filesystems/transfers/%s", ETCD_PREFIX, args.TransferRequestId,
+		),
+		string(serialized),
+		nil,
+	)
+	if err != nil {
+		return err
+	}
+	// XXX A transfer should be able to span multiple filesystemIds, really. So
+	// tying a transfer to a filesystem id is probably wrong. except, the thing
+	// being updated is a specific branch (filesystem id), it's ok if it drags
+	// dependent snapshots along with it.
+	responseChan, err := d.state.globalFsRequest(args.FilesystemId, &Event{
+		Name: "peer-transfer",
+		Args: &EventArgs{
+			"Transfer": args,
+		},
+	})
+	if err != nil {
+		return err
+	}
+
+	// Block until the fsmachine is ready to transfer
+	log.Printf("[RegisterTransfer:%s] waiting for ack from the fsmachine...", args.FilesystemId)
+	e := <-responseChan
+	log.Printf("[RegisterTransfer:%s] received ack from the fsmachine: %+v", args.FilesystemId, e)
+
+	if e.Name != "awaiting-transfer" {
+		// Something went wrong!
+		return fmt.Errorf("Error requesting peer transfer: %+v", e)
+	} else {
+		return nil
+	}
+
+	return nil
+}
+
 // Need both push and pull because one cluster will often be behind NAT.
 // Transfer will immediately return a transferId which can be queried until
 // completion