forked from moby/swarmkit
-
Notifications
You must be signed in to change notification settings - Fork 0
/
testutils.go
482 lines (425 loc) · 13.2 KB
/
testutils.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
package testutils
import (
"errors"
"fmt"
"io/ioutil"
"net"
"os"
"reflect"
"testing"
"time"
"golang.org/x/net/context"
"google.golang.org/grpc"
etcdraft "github.com/coreos/etcd/raft"
"github.com/docker/swarmkit/api"
"github.com/docker/swarmkit/ca"
cautils "github.com/docker/swarmkit/ca/testutils"
"github.com/docker/swarmkit/identity"
"github.com/docker/swarmkit/manager/state/raft"
"github.com/docker/swarmkit/manager/state/store"
"github.com/pivotal-golang/clock/fakeclock"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// TestNode represents a raft test node
type TestNode struct {
*raft.Node
Listener *wrappedListener
SecurityConfig *ca.SecurityConfig
}
// AdvanceTicks advances the raft state machine fake clock
func AdvanceTicks(clockSource *fakeclock.FakeClock, ticks int) {
// A FakeClock timer won't fire multiple times if time is advanced
// more than its interval.
for i := 0; i != ticks; i++ {
clockSource.Increment(time.Second)
}
}
// PollFuncWithTimeout is used to periodically execute a check function, it
// returns error after timeout.
func PollFuncWithTimeout(clockSource *fakeclock.FakeClock, f func() error, timeout time.Duration) error {
if f() == nil {
return nil
}
timer := time.After(timeout)
for {
if clockSource != nil {
clockSource.Increment(time.Second)
}
err := f()
if err == nil {
return nil
}
select {
case <-timer:
return fmt.Errorf("polling failed: %v", err)
case <-time.After(50 * time.Millisecond):
}
}
}
// PollFunc is like PollFuncWithTimeout with timeout=10s.
func PollFunc(clockSource *fakeclock.FakeClock, f func() error) error {
return PollFuncWithTimeout(clockSource, f, 10*time.Second)
}
// WaitForCluster waits until leader will be one of specified nodes
func WaitForCluster(t *testing.T, clockSource *fakeclock.FakeClock, nodes map[uint64]*TestNode) {
err := PollFunc(clockSource, func() error {
var prev *etcdraft.Status
nodeLoop:
for _, n := range nodes {
if prev == nil {
prev = new(etcdraft.Status)
*prev = n.Status()
for _, n2 := range nodes {
if n2.Config.ID == prev.Lead {
continue nodeLoop
}
}
return errors.New("did not find leader in member list")
}
cur := n.Status()
for _, n2 := range nodes {
if n2.Config.ID == cur.Lead {
if cur.Lead != prev.Lead || cur.Term != prev.Term || cur.Applied != prev.Applied {
return errors.New("state does not match on all nodes")
}
continue nodeLoop
}
}
return errors.New("did not find leader in member list")
}
return nil
})
require.NoError(t, err)
}
// WaitForPeerNumber waits until peers in cluster converge to specified number
func WaitForPeerNumber(t *testing.T, clockSource *fakeclock.FakeClock, nodes map[uint64]*TestNode, count int) {
assert.NoError(t, PollFunc(clockSource, func() error {
for _, n := range nodes {
if len(n.GetMemberlist()) != count {
return errors.New("unexpected number of members")
}
}
return nil
}))
}
// wrappedListener disables the Close method to make it possible to reuse a
// socket. close must be called to release the socket.
type wrappedListener struct {
net.Listener
acceptConn chan net.Conn
acceptErr chan error
closed chan struct{}
}
func newWrappedListener(l net.Listener) *wrappedListener {
wrappedListener := wrappedListener{
Listener: l,
acceptConn: make(chan net.Conn),
acceptErr: make(chan error, 1),
closed: make(chan struct{}, 10), // grpc closes multiple times
}
// Accept connections
go func() {
for {
conn, err := l.Accept()
if err != nil {
wrappedListener.acceptErr <- err
return
}
wrappedListener.acceptConn <- conn
}
}()
return &wrappedListener
}
func (l *wrappedListener) Accept() (net.Conn, error) {
// closure must take precendence over taking a connection
// from the channel
select {
case <-l.closed:
return nil, errors.New("listener closed")
default:
}
select {
case conn := <-l.acceptConn:
return conn, nil
case err := <-l.acceptErr:
return nil, err
case <-l.closed:
return nil, errors.New("listener closed")
}
}
func (l *wrappedListener) Close() error {
l.closed <- struct{}{}
return nil
}
func (l *wrappedListener) close() error {
return l.Listener.Close()
}
// recycleWrappedListener creates a new wrappedListener that uses the same
// listening socket as the supplied wrappedListener.
func recycleWrappedListener(old *wrappedListener) *wrappedListener {
return &wrappedListener{
Listener: old.Listener,
acceptConn: old.acceptConn,
acceptErr: old.acceptErr,
closed: make(chan struct{}, 10), // grpc closes multiple times
}
}
// NewNode creates a new raft node to use for tests
func NewNode(t *testing.T, clockSource *fakeclock.FakeClock, tc *cautils.TestCA, opts ...raft.NewNodeOptions) *TestNode {
l, err := net.Listen("tcp", "127.0.0.1:0")
require.NoError(t, err, "can't bind to raft service port")
wrappedListener := newWrappedListener(l)
securityConfig, err := tc.NewNodeConfig(ca.ManagerRole)
require.NoError(t, err)
serverOpts := []grpc.ServerOption{grpc.Creds(securityConfig.ServerTLSCreds)}
s := grpc.NewServer(serverOpts...)
cfg := raft.DefaultNodeConfig()
stateDir, err := ioutil.TempDir("", "test-raft")
require.NoError(t, err, "can't create temporary state directory")
newNodeOpts := raft.NewNodeOptions{
ID: securityConfig.ClientTLSCreds.NodeID(),
Addr: l.Addr().String(),
Config: cfg,
StateDir: stateDir,
ClockSource: clockSource,
SendTimeout: 10 * time.Second,
TLSCredentials: securityConfig.ClientTLSCreds,
}
if len(opts) > 1 {
panic("more than one optional argument provided")
}
if len(opts) == 1 {
newNodeOpts.JoinAddr = opts[0].JoinAddr
}
n, err := raft.NewNode(context.Background(), newNodeOpts)
require.NoError(t, err, "can't create raft node")
n.Server = s
return &TestNode{Node: n, Listener: wrappedListener, SecurityConfig: securityConfig}
}
// NewInitNode creates a new raft node initiating the cluster
// for other members to join
func NewInitNode(t *testing.T, tc *cautils.TestCA, raftConfig *api.RaftConfig, opts ...raft.NewNodeOptions) (*TestNode, *fakeclock.FakeClock) {
ctx := context.Background()
clockSource := fakeclock.NewFakeClock(time.Now())
n := NewNode(t, clockSource, tc, opts...)
go n.Run(ctx)
if raftConfig != nil {
assert.NoError(t, n.MemoryStore().Update(func(tx store.Tx) error {
return store.CreateCluster(tx, &api.Cluster{
ID: identity.NewID(),
Spec: api.ClusterSpec{
Annotations: api.Annotations{
Name: store.DefaultClusterName,
},
Raft: *raftConfig,
},
})
}))
}
raft.Register(n.Server, n.Node)
go func() {
// After stopping, we should receive an error from Serve
assert.Error(t, n.Server.Serve(n.Listener))
}()
return n, clockSource
}
// NewJoinNode creates a new raft node joining an existing cluster
func NewJoinNode(t *testing.T, clockSource *fakeclock.FakeClock, join string, tc *cautils.TestCA, opts ...raft.NewNodeOptions) *TestNode {
var derivedOpts raft.NewNodeOptions
if len(opts) == 1 {
derivedOpts = opts[0]
}
derivedOpts.JoinAddr = join
n := NewNode(t, clockSource, tc, derivedOpts)
go n.Run(context.Background())
raft.Register(n.Server, n.Node)
go func() {
// After stopping, we should receive an error from Serve
assert.Error(t, n.Server.Serve(n.Listener))
}()
return n
}
// RestartNode restarts a raft test node
func RestartNode(t *testing.T, clockSource *fakeclock.FakeClock, oldNode *TestNode, forceNewCluster bool) *TestNode {
wrappedListener := recycleWrappedListener(oldNode.Listener)
securityConfig := oldNode.SecurityConfig
serverOpts := []grpc.ServerOption{grpc.Creds(securityConfig.ServerTLSCreds)}
s := grpc.NewServer(serverOpts...)
cfg := raft.DefaultNodeConfig()
newNodeOpts := raft.NewNodeOptions{
ID: securityConfig.ClientTLSCreds.NodeID(),
Addr: oldNode.Address,
Config: cfg,
StateDir: oldNode.StateDir,
ForceNewCluster: forceNewCluster,
ClockSource: clockSource,
SendTimeout: 10 * time.Second,
TLSCredentials: securityConfig.ClientTLSCreds,
}
ctx := context.Background()
n, err := raft.NewNode(ctx, newNodeOpts)
require.NoError(t, err, "can't create raft node")
n.Server = s
go n.Run(ctx)
raft.Register(s, n)
go func() {
// After stopping, we should receive an error from Serve
assert.Error(t, s.Serve(wrappedListener))
}()
return &TestNode{Node: n, Listener: wrappedListener, SecurityConfig: securityConfig}
}
// NewRaftCluster creates a new raft cluster with 3 nodes for testing
func NewRaftCluster(t *testing.T, tc *cautils.TestCA, config ...*api.RaftConfig) (map[uint64]*TestNode, *fakeclock.FakeClock) {
var (
raftConfig *api.RaftConfig
clockSource *fakeclock.FakeClock
)
if len(config) > 1 {
panic("more than one optional argument provided")
}
if len(config) == 1 {
raftConfig = config[0]
}
nodes := make(map[uint64]*TestNode)
nodes[1], clockSource = NewInitNode(t, tc, raftConfig)
AddRaftNode(t, clockSource, nodes, tc)
AddRaftNode(t, clockSource, nodes, tc)
return nodes, clockSource
}
// AddRaftNode adds an additional raft test node to an existing cluster
func AddRaftNode(t *testing.T, clockSource *fakeclock.FakeClock, nodes map[uint64]*TestNode, tc *cautils.TestCA, opts ...raft.NewNodeOptions) {
n := uint64(len(nodes) + 1)
nodes[n] = NewJoinNode(t, clockSource, nodes[1].Address, tc, opts...)
WaitForCluster(t, clockSource, nodes)
}
// TeardownCluster destroys a raft cluster used for tests
func TeardownCluster(t *testing.T, nodes map[uint64]*TestNode) {
for _, node := range nodes {
ShutdownNode(node)
node.Listener.close()
}
}
// ShutdownNode shuts down a raft test node and deletes the content
// of the state directory
func ShutdownNode(node *TestNode) {
node.Server.Stop()
node.Shutdown()
os.RemoveAll(node.StateDir)
}
// Leader determines who is the leader amongst a set of raft nodes
// belonging to the same cluster
func Leader(nodes map[uint64]*TestNode) *TestNode {
for _, n := range nodes {
if n.Config.ID == n.Leader() {
return n
}
}
panic("could not find a leader")
}
// ProposeValue proposes a value to a raft test cluster
func ProposeValue(t *testing.T, raftNode *TestNode, nodeID ...string) (*api.Node, error) {
nodeIDStr := "id1"
if len(nodeID) != 0 {
nodeIDStr = nodeID[0]
}
node := &api.Node{
ID: nodeIDStr,
Spec: api.NodeSpec{
Annotations: api.Annotations{
Name: nodeIDStr,
},
},
}
storeActions := []*api.StoreAction{
{
Action: api.StoreActionKindCreate,
Target: &api.StoreAction_Node{
Node: node,
},
},
}
ctx, _ := context.WithTimeout(context.Background(), time.Second)
err := raftNode.ProposeValue(ctx, storeActions, func() {
err := raftNode.MemoryStore().ApplyStoreActions(storeActions)
assert.NoError(t, err, "error applying actions")
})
if err != nil {
return nil, err
}
return node, nil
}
// CheckValue checks that the value has been propagated between raft members
func CheckValue(t *testing.T, clockSource *fakeclock.FakeClock, raftNode *TestNode, createdNode *api.Node) {
assert.NoError(t, PollFunc(clockSource, func() error {
var err error
raftNode.MemoryStore().View(func(tx store.ReadTx) {
var allNodes []*api.Node
allNodes, err = store.FindNodes(tx, store.All)
if err != nil {
return
}
if len(allNodes) != 1 {
err = fmt.Errorf("expected 1 node, got %d nodes", len(allNodes))
return
}
if !reflect.DeepEqual(allNodes[0], createdNode) {
err = errors.New("node did not match expected value")
}
})
return err
}))
}
// CheckNoValue checks that there is no value replicated on nodes, generally
// used to test the absence of a leader
func CheckNoValue(t *testing.T, clockSource *fakeclock.FakeClock, raftNode *TestNode) {
assert.NoError(t, PollFunc(clockSource, func() error {
var err error
raftNode.MemoryStore().View(func(tx store.ReadTx) {
var allNodes []*api.Node
allNodes, err = store.FindNodes(tx, store.All)
if err != nil {
return
}
if len(allNodes) != 0 {
err = fmt.Errorf("expected no nodes, got %d", len(allNodes))
}
})
return err
}))
}
// CheckValuesOnNodes checks that all the nodes in the cluster have the same
// replicated data, generally used to check if a node can catch up with the logs
// correctly
func CheckValuesOnNodes(t *testing.T, clockSource *fakeclock.FakeClock, checkNodes map[uint64]*TestNode, ids []string, values []*api.Node) {
iteration := 0
for checkNodeID, node := range checkNodes {
assert.NoError(t, PollFunc(clockSource, func() error {
var err error
node.MemoryStore().View(func(tx store.ReadTx) {
var allNodes []*api.Node
allNodes, err = store.FindNodes(tx, store.All)
if err != nil {
return
}
for i, id := range ids {
n := store.GetNode(tx, id)
if n == nil {
err = fmt.Errorf("node %s not found on %d (iteration %d)", id, checkNodeID, iteration)
return
}
if !reflect.DeepEqual(values[i], n) {
err = fmt.Errorf("node %s did not match expected value on %d (iteration %d)", id, checkNodeID, iteration)
return
}
}
if len(allNodes) != len(ids) {
err = fmt.Errorf("expected %d nodes, got %d (iteration %d)", len(ids), len(allNodes), iteration)
return
}
})
return err
}))
iteration++
}
}