allocs: Add nomad alloc stop

This adds a `nomad alloc stop` command that can be used to stop and
force migrate an allocation to a different node.

This is built on top of the AllocUpdateDesiredTransitionRequest and
explicitly limits the scope of access to that transition to expose it
under the alloc-lifecycle ACL.

The API returns the follow up eval that can be used as part of
monitoring in the CLI or parsed and used in an external tool.

api/allocations.go

@@ -89,6 +89,20 @@ func (a *Allocations) Restart(alloc *Allocation, taskName string, q *QueryOption
 	return err
 }
 
+func (a *Allocations) Stop(alloc *Allocation, q *QueryOptions) (*AllocStopResponse, error) {
+	var resp AllocStopResponse
+	_, err := a.client.putQuery("/v1/allocation/"+alloc.ID+"/stop", nil, &resp, q)
+	return &resp, err
+}
+
+// AllocStopResponse is the response to an `AllocStopRequest`
+type AllocStopResponse struct {
+	// EvalID is the id of the follow up evalution for the rescheduled alloc.
+	EvalID string
+
+	WriteMeta
+}
+
 // Allocation is used for serialization of allocations.
 type Allocation struct {
 	ID                    string

command/agent/alloc_endpoint.go

@@ -42,7 +42,29 @@ func (s *HTTPServer) AllocsRequest(resp http.ResponseWriter, req *http.Request)
 }
 
 func (s *HTTPServer) AllocSpecificRequest(resp http.ResponseWriter, req *http.Request) (interface{}, error) {
-	allocID := strings.TrimPrefix(req.URL.Path, "/v1/allocation/")
+	reqSuffix := strings.TrimPrefix(req.URL.Path, "/v1/allocation/")
+
+	// tokenize the suffix of the path to get the alloc id and find the action
+	// invoked on the alloc id
+	tokens := strings.Split(reqSuffix, "/")
+	if len(tokens) > 2 || len(tokens) < 1 {
+		return nil, CodedError(404, resourceNotFoundErr)
+	}
+	allocID := tokens[0]
+
+	if len(tokens) == 1 {
+		return s.allocGet(allocID, resp, req)
+	}
+
+	switch tokens[1] {
+	case "stop":
+		return s.allocStop(allocID, resp, req)
+	}
+
+	return nil, CodedError(404, resourceNotFoundErr)
+}
+
+func (s *HTTPServer) allocGet(allocID string, resp http.ResponseWriter, req *http.Request) (interface{}, error) {
 	if req.Method != "GET" {
 		return nil, CodedError(405, ErrInvalidMethod)
 	}
@@ -79,8 +101,22 @@ func (s *HTTPServer) AllocSpecificRequest(resp http.ResponseWriter, req *http.Re
 	return alloc, nil
 }
 
-func (s *HTTPServer) ClientAllocRequest(resp http.ResponseWriter, req *http.Request) (interface{}, error) {
+func (s *HTTPServer) allocStop(allocID string, resp http.ResponseWriter, req *http.Request) (interface{}, error) {
+	if !(req.Method == "POST" || req.Method == "PUT") {
+		return nil, CodedError(405, ErrInvalidMethod)
+	}
 
+	sr := &structs.AllocStopRequest{
+		AllocID: allocID,
+	}
+	s.parseWriteRequest(req, &sr.WriteRequest)
+
+	var out structs.AllocStopResponse
+	err := s.agent.RPC("Alloc.Stop", &sr, &out)
+	return &out, err
+}
+
+func (s *HTTPServer) ClientAllocRequest(resp http.ResponseWriter, req *http.Request) (interface{}, error) {
 	reqSuffix := strings.TrimPrefix(req.URL.Path, "/v1/client/allocation/")
 
 	// tokenize the suffix of the path to get the alloc id and find the action

command/agent/alloc_endpoint_test.go

@@ -394,6 +394,49 @@ func TestHTTP_AllocRestart_ACL(t *testing.T) {
 	})
 }
 
+func TestHTTP_AllocStop(t *testing.T) {
+	t.Parallel()
+	httpTest(t, nil, func(s *TestAgent) {
+		// Directly manipulate the state
+		state := s.Agent.server.State()
+		alloc := mock.Alloc()
+		require := require.New(t)
+		require.NoError(state.UpsertJobSummary(999, mock.JobSummary(alloc.JobID)))
+
+		require.NoError(state.UpsertAllocs(1000, []*structs.Allocation{alloc}))
+
+		{
+			// Make the HTTP request
+			req, err := http.NewRequest("POST", "/v1/allocation/"+alloc.ID+"/stop", nil)
+			require.NoError(err)
+			respW := httptest.NewRecorder()
+
+			// Make the request
+			obj, err := s.Server.AllocSpecificRequest(respW, req)
+			require.NoError(err)
+
+			a := obj.(*structs.AllocStopResponse)
+			require.NotEmpty(a.EvalID, "missing eval")
+			require.NotEmpty(a.Index, "missing index")
+		}
+
+		{
+			// Make the HTTP request
+			req, err := http.NewRequest("POST", "/v1/allocation/"+alloc.ID+"/stop", nil)
+			require.NoError(err)
+			respW := httptest.NewRecorder()
+
+			// Make the request
+			obj, err := s.Server.AllocSpecificRequest(respW, req)
+			require.NoError(err)
+
+			a := obj.(*structs.AllocStopResponse)
+			require.NotEmpty(a.EvalID, "missing eval")
+			require.NotEmpty(a.Index, "missing index")
+		}
+	})
+}
+
 func TestHTTP_AllocStats(t *testing.T) {
 	t.Parallel()
 	require := require.New(t)

command/alloc_stop.go

@@ -0,0 +1,128 @@
+package command
+
+import (
+	"fmt"
+	"strings"
+)
+
+type AllocStopCommand struct {
+	Meta
+}
+
+func (a *AllocStopCommand) Help() string {
+	helpText := `
+Usage: nomad alloc stop [options] <allocation>
+Alias: nomad stop
+
+  stop an existing allocation. This command is used to signal a specific alloc
+  to shut down. When the allocation has been shut down, it will then be
+  rescheduled. An interactive monitoring session will display log lines as the
+  allocation completes shutting down. It is safe to exit the monitor early with
+  ctrl-c.
+
+General Options:
+
+  ` + generalOptionsUsage() + `
+
+Stop Specific Options:
+
+  -detach
+    Return immediately instead of entering monitor mode. After the
+    stop command is submitted, a new evaluation ID is printed to the
+    screen, which can be used to examine the rescheduling evaluation using the
+    eval-status command.
+
+  -verbose
+    Show full information.
+`
+	return strings.TrimSpace(helpText)
+}
+
+func (c *AllocStopCommand) Name() string { return "alloc stop" }
+
+func (c *AllocStopCommand) Run(args []string) int {
+	var detach, verbose bool
+
+	flags := c.Meta.FlagSet(c.Name(), FlagSetClient)
+	flags.Usage = func() { c.Ui.Output(c.Help()) }
+	flags.BoolVar(&detach, "detach", false, "")
+	flags.BoolVar(&verbose, "verbose", false, "")
+
+	if err := flags.Parse(args); err != nil {
+		return 1
+	}
+
+	// Check that we got exactly one alloc
+	args = flags.Args()
+	if len(args) != 1 {
+		c.Ui.Error("This command takes one argument: <alloc-id>")
+		c.Ui.Error(commandErrorText(c))
+		return 1
+	}
+
+	allocID := args[0]
+
+	// Truncate the id unless full length is requested
+	length := shortId
+	if verbose {
+		length = fullId
+	}
+
+	// Query the allocation info
+	if len(allocID) == 1 {
+		c.Ui.Error(fmt.Sprintf("Alloc ID must contain at least two characters."))
+		return 1
+	}
+
+	allocID = sanitizeUUIDPrefix(allocID)
+
+	// Get the HTTP client
+	client, err := c.Meta.Client()
+	if err != nil {
+		c.Ui.Error(fmt.Sprintf("Error initializing client: %s", err))
+		return 1
+	}
+
+	allocs, _, err := client.Allocations().PrefixList(allocID)
+	if err != nil {
+		c.Ui.Error(fmt.Sprintf("Error querying allocation: %v", err))
+		return 1
+	}
+
+	if len(allocs) == 0 {
+		c.Ui.Error(fmt.Sprintf("No allocation(s) with prefix or id %q found", allocID))
+		return 1
+	}
+
+	if len(allocs) > 1 {
+		// Format the allocs
+		out := formatAllocListStubs(allocs, verbose, length)
+		c.Ui.Error(fmt.Sprintf("Prefix matched multiple allocations\n\n%s", out))
+		return 1
+	}
+
+	// Prefix lookup matched a single allocation
+	alloc, _, err := client.Allocations().Info(allocs[0].ID, nil)
+	if err != nil {
+		c.Ui.Error(fmt.Sprintf("Error querying allocation: %s", err))
+		return 1
+	}
+
+	resp, err := client.Allocations().Stop(alloc, nil)
+	if err != nil {
+		c.Ui.Error(fmt.Sprintf("Error stopping allocation: %s", err))
+		return 1
+	}
+
+	if detach {
+		c.Ui.Output(resp.EvalID)
+		return 0
+	}
+
+	mon := newMonitor(c.Ui, client, length)
+	return mon.monitor(resp.EvalID, false)
+}
+
+func (a *AllocStopCommand) Synopsis() string {
+	return "Stop and reschedule a running allocation"
+}

command/alloc_stop_test.go

@@ -0,0 +1,112 @@
+package command
+
+import (
+	"fmt"
+	"testing"
+
+	"github.com/hashicorp/nomad/api"
+	"github.com/hashicorp/nomad/nomad/structs"
+	"github.com/hashicorp/nomad/testutil"
+	"github.com/mitchellh/cli"
+	"github.com/stretchr/testify/require"
+)
+
+func TestAllocStopCommand_Implements(t *testing.T) {
+	t.Parallel()
+	var _ cli.Command = &AllocStopCommand{}
+}
+
+func TestAllocStop_Fails(t *testing.T) {
+	srv, _, url := testServer(t, false, nil)
+	defer srv.Shutdown()
+
+	require := require.New(t)
+	ui := new(cli.MockUi)
+	cmd := &AllocStopCommand{Meta: Meta{Ui: ui}}
+
+	// Fails on misuse
+	require.Equal(cmd.Run([]string{"some", "garbage", "args"}), 1, "Expected failure")
+	require.Contains(ui.ErrorWriter.String(), commandErrorText(cmd), "Expected help output")
+	ui.ErrorWriter.Reset()
+
+	// Fails on connection failure
+	require.Equal(cmd.Run([]string{"-address=nope", "foobar"}), 1, "expected failure")
+	require.Contains(ui.ErrorWriter.String(), "Error querying allocation")
+	ui.ErrorWriter.Reset()
+
+	// Fails on missing alloc
+	require.Equal(cmd.Run([]string{"-address=" + url, "26470238-5CF2-438F-8772-DC67CFB0705C"}), 1)
+	require.Contains(ui.ErrorWriter.String(), "No allocation(s) with prefix or id")
+	ui.ErrorWriter.Reset()
+
+	// Fail on identifier with too few characters
+	require.Equal(cmd.Run([]string{"-address=" + url, "2"}), 1)
+	require.Contains(ui.ErrorWriter.String(), "must contain at least two characters")
+	ui.ErrorWriter.Reset()
+
+	// Identifiers with uneven length should produce a query result
+	require.Equal(cmd.Run([]string{"-address=" + url, "123"}), 1)
+	require.Contains(ui.ErrorWriter.String(), "No allocation(s) with prefix or id")
+	ui.ErrorWriter.Reset()
+}
+
+func TestAllocStop_Run(t *testing.T) {
+	srv, client, url := testServer(t, true, nil)
+	defer srv.Shutdown()
+
+	require := require.New(t)
+
+	// Wait for a node to be ready
+	testutil.WaitForResult(func() (bool, error) {
+		nodes, _, err := client.Nodes().List(nil)
+		if err != nil {
+			return false, err
+		}
+		for _, node := range nodes {
+			if _, ok := node.Drivers["mock_driver"]; ok &&
+				node.Status == structs.NodeStatusReady {
+				return true, nil
+			}
+		}
+		return false, fmt.Errorf("no ready nodes")
+	}, func(err error) {
+		t.Fatalf("err: %v", err)
+	})
+
+	ui := new(cli.MockUi)
+	cmd := &AllocStopCommand{Meta: Meta{Ui: ui}}
+
+	jobID := "job1_sfx"
+	job1 := testJob(jobID)
+	resp, _, err := client.Jobs().Register(job1, nil)
+	require.NoError(err)
+	if code := waitForSuccess(ui, client, fullId, t, resp.EvalID); code != 0 {
+		t.Fatalf("status code non zero saw %d", code)
+	}
+	// get an alloc id
+	allocId1 := ""
+	if allocs, _, err := client.Jobs().Allocations(jobID, false, nil); err == nil {
+		if len(allocs) > 0 {
+			allocId1 = allocs[0].ID
+		}
+	}
+	require.NotEmpty(allocId1, "unable to find allocation")
+
+	// Wait for alloc to be running
+	testutil.WaitForResult(func() (bool, error) {
+		alloc, _, err := client.Allocations().Info(allocId1, nil)
+		if err != nil {
+			return false, err
+		}
+		if alloc.ClientStatus == api.AllocClientStatusRunning {
+			return true, nil
+		}
+		return false, fmt.Errorf("alloc is not running, is: %s", alloc.ClientStatus)
+	}, func(err error) {
+		t.Fatalf("err: %v", err)
+	})
+
+	require.Equal(cmd.Run([]string{"-address=" + url, allocId1}), 0, "expected successful exit code")
+
+	ui.OutputWriter.Reset()
+}

command/commands.go

@@ -140,6 +140,11 @@ func Commands(metaPtr *Meta, agentUi cli.Ui) map[string]cli.CommandFactory {
 				Meta: meta,
 			}, nil
 		},
+		"alloc stop": func() (cli.Command, error) {
+			return &AllocStopCommand{
+				Meta: meta,
+			}, nil
+		},
 		"alloc fs": func() (cli.Command, error) {
 			return &AllocFSCommand{
 				Meta: meta,