-
Notifications
You must be signed in to change notification settings - Fork 914
/
orchestrator.go
640 lines (511 loc) · 16.3 KB
/
orchestrator.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
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
package orchestrator
import (
"fmt"
"net"
"sync"
"time"
"github.com/botlabs-gg/yagpdb/v2/lib/discordgo"
"github.com/botlabs-gg/yagpdb/v2/lib/dshardorchestrator"
"github.com/pkg/errors"
)
type NodeIDProvider interface {
GenerateID() string
}
// RecommendTotalShardCountProvider will only be called when a fresh new shard count is needed
// this is only the case if: no nodes were running with any shards 10 seconds after the orchestrator starts up
//
// if new nodes without shards connect during this period, they will be put on hold while waiting for nodes with
// running shards to re-identify with the orchestrator, thereby keeping the total shard count across restarts of the orchestrator
// in a fairly realiable manner
//
// but using this interface you can implement completely fine grained control (say storing the shard count in persistent store and updating it manually)
type RecommendTotalShardCountProvider interface {
GetTotalShardCount() (int, error)
}
// NodeLauncher is responsible for the logic of spinning up new processes and also receiving the version of the node that would be launched
type NodeLauncher interface {
// Launches a new node, returning the id and a error if something went wrong
LaunchNewNode() (nodeID string, err error)
// Retrieves the version of nodes we would launch if we were to call LaunchNewNode()
// for example, the vesion of the binary deployed on the server.
LaunchVersion() (version string, err error)
}
// VersionUpdater is reponsible for updating the deployment, for example pulling a new version from a CI server
type VersionUpdater interface {
PullNewVersion() (newVersion string, err error)
}
type Orchestrator struct {
// these fields are only safe to edit before you start the orchestrator
// if you decide to change anything afterwards, it may panic or cause undefined behaviour
ShardCountProvider RecommendTotalShardCountProvider
NodeLauncher NodeLauncher
Logger dshardorchestrator.Logger
VersionUpdater VersionUpdater
// this is for running in a multi-host mode
// this allows you to have 1 shard orchestrator per host, then only have that orchestator care about the specified shards
// this also requires that the total shard count is fixed.
// FixedTotalShardCount will be ignored if <1
// ResponsibleForShards will be ignored if len < 1
FixedTotalShardCount int
ResponsibleForShards []int
// if set, the orchestrator will make sure that all the shards are always running
EnsureAllShardsRunning bool
// For large bot sharding the bucket size should be 16
// the orchestrator will only put shards in the same (bucket/bucketspernode) on the same node
ShardBucketSize int
// The number of buckets per node, this * shardBucketSize should equal to the actual bots bucket size, but this allows more gradual control of the startup process
BucketsPerNode int
// the max amount of downtime for a node before we consider it dead and it will start a new node for those shards
// if set to below zero then it will not perform the restart at all
MaxNodeDowntimeBeforeRestart time.Duration
// the maximum amount of shards per node, note that this is solely for the automated tasks the orchestrator provides
// and you can still go over it if you manually start shards on a node
MaxShardsPerNode int
// in case we are intiailizing max shards from nodes, we wait 10 seconds when we start before we decide we need to fetch a fresh shard count
SkipSafeStartupDelayMaxShards bool
monitor *monitor
// below fields are protected by the following mutex
mu sync.Mutex
connectedNodes []*NodeConn
totalShards int
activeMigrationFrom string
activeMigrationTo string
netListener net.Listener
// blacklisted nodes will not get new shards assigned to them
blacklistedNodes []string
performingFullMigration bool
}
func NewStandardOrchestrator(session *discordgo.Session) *Orchestrator {
return &Orchestrator{
ShardCountProvider: &StdShardCountProvider{DiscordSession: session},
}
}
// Start will start the orchestrator, and start to listen for clients on the specified address
// IMPORTANT: opening this up to the outer internet is bad because there's no authentication.
func (o *Orchestrator) Start(listenAddr string) error {
o.mu.Lock()
defer o.mu.Unlock()
err := o.openListen(listenAddr)
if err != nil {
return err
}
o.monitor = &monitor{
orchestrator: o,
stopChan: make(chan bool),
}
go o.monitor.run()
return nil
}
// Stop will stop the orchestrator and the monitor
func (o *Orchestrator) Stop() {
o.mu.Lock()
defer o.mu.Unlock()
o.netListener.Close()
o.monitor.stop()
}
// openListen starts listening for slave connections on the specified address
func (o *Orchestrator) openListen(addr string) error {
listener, err := net.Listen("tcp", addr)
if err != nil {
return errors.WithMessage(err, "net.Listen")
}
o.netListener = listener
// go monitorSlaves()
go o.listenForNodes(listener)
return nil
}
func (o *Orchestrator) listenForNodes(listener net.Listener) {
o.Log(dshardorchestrator.LogInfo, nil, "listening for incoming nodes on: "+listener.Addr().String())
for {
conn, err := listener.Accept()
if err != nil {
o.Log(dshardorchestrator.LogError, err, "failed accepting incmoing connection")
break
}
o.Log(dshardorchestrator.LogInfo, nil, "new node connection!")
client := o.NewNodeConn(conn)
o.mu.Lock()
o.connectedNodes = append(o.connectedNodes, client)
o.mu.Unlock()
go client.listen()
}
}
func (o *Orchestrator) FindNodeByID(id string) *NodeConn {
o.mu.Lock()
for _, v := range o.connectedNodes {
if v.Conn.GetID() == id {
o.mu.Unlock()
return v
}
}
o.mu.Unlock()
return nil
}
type NodeStatus struct {
ID string
Version string
SessionEstablished bool
Shards []int
Connected bool
DisconnectedAt time.Time
Blacklisted bool
MigratingFrom string
MigratingTo string
MigratingShard int
}
// GetFullNodesStatus returns the full status of all nodes
func (o *Orchestrator) GetFullNodesStatus() []*NodeStatus {
result := make([]*NodeStatus, 0)
o.mu.Lock()
for _, v := range o.connectedNodes {
o.mu.Unlock()
st := v.GetFullStatus()
o.mu.Lock()
st.Blacklisted = o.isNodeBlacklisted(false, st.ID)
result = append(result, st)
}
o.mu.Unlock()
return result
}
var (
ErrUnknownFromNode = errors.New("unknown 'from' node")
ErrUnknownToNode = errors.New("unknown 'to' node")
ErrFromNodeNotRunningShard = errors.New("'from' node not running shard")
ErrNodeBusy = errors.New("node is busy")
)
// StartShardMigration attempts to start a shard migration, moving shardID from a origin node to a destination node
func (o *Orchestrator) StartShardMigration(toNodeID string, shardID int) error {
// find the origin node
fromNodeID := ""
status := o.GetFullNodesStatus()
OUTER:
for _, n := range status {
if !n.Connected {
continue
}
for _, s := range n.Shards {
if s == shardID {
fromNodeID = n.ID
break OUTER
}
}
}
toNode := o.FindNodeByID(toNodeID)
fromNode := o.FindNodeByID(fromNodeID)
if fromNode == nil {
return ErrUnknownFromNode
}
if toNode == nil {
return ErrUnknownToNode
}
// mark the origin node as busy
fromNode.mu.Lock()
// make sure the node were migrating from actually holds the shard
foundShard := false
for _, v := range fromNode.runningShards {
if v == shardID {
foundShard = true
break
}
}
// it did not hold the shard
if !foundShard {
fromNode.mu.Unlock()
return ErrFromNodeNotRunningShard
}
// make sure its not busy
if fromNode.shardMigrationMode != dshardorchestrator.ShardMigrationModeNone {
fromNode.mu.Unlock()
return ErrNodeBusy
}
fromNode.shardMigrationMode = dshardorchestrator.ShardMigrationModeFrom
fromNode.shardMigrationOtherNodeID = toNodeID
fromNode.shardMigrationShard = shardID
// fromNode.shardmigrationTotalUserEvts = 0
fromNode.mu.Unlock()
// mark the destination node as busy
toNode.mu.Lock()
// make sure its not busy
if toNode.shardMigrationMode != dshardorchestrator.ShardMigrationModeNone {
toNode.mu.Unlock()
fromNode.mu.Lock()
// need to rollback the from state as we cannot go further
fromNode.shardMigrationMode = dshardorchestrator.ShardMigrationModeNone
fromNode.shardMigrationOtherNodeID = ""
fromNode.shardMigrationShard = -1
fromNode.mu.Unlock()
toNode.mu.Unlock()
return ErrNodeBusy
}
toNode.shardMigrationMode = dshardorchestrator.ShardMigrationModeTo
toNode.shardMigrationOtherNodeID = fromNodeID
toNode.shardMigrationShard = shardID
toNode.mu.Unlock()
o.Log(dshardorchestrator.LogInfo, nil, fmt.Sprintf("migrating shard %d from %q to %q", shardID, fromNodeID, toNodeID))
// everything passed, we can start the migration of the shard
fromNode.Conn.SendLogErr(dshardorchestrator.EvtPrepareShardmigration, &dshardorchestrator.PrepareShardmigrationData{
Origin: true,
ShardID: shardID,
})
return nil
}
// Log will log to the designated logger or he standard logger
func (o *Orchestrator) Log(level dshardorchestrator.LogLevel, err error, msg string) {
if err != nil {
msg = msg + ": " + err.Error()
}
if o.Logger == nil {
dshardorchestrator.StdLogInstance.Log(level, msg)
} else {
o.Logger.Log(level, msg)
}
}
var (
ErrNoNodeLauncher = errors.New("orchestrator.NodeLauncher is nil")
)
// StartNewNode will launch a new node, it will not wait for it to connect
func (o *Orchestrator) StartNewNode() (string, error) {
if o.NodeLauncher == nil {
return "", ErrNoNodeLauncher
}
return o.NodeLauncher.LaunchNewNode()
}
var (
ErrShardAlreadyRunning = errors.New("shard already running")
ErrUnknownNode = errors.New("unknown node")
)
// StartShard will start the specified shard on the specified node
// it will return ErrShardAlreadyRunning if the shard is running on another node already
func (o *Orchestrator) StartShards(nodeID string, shards ...int) error {
fullStatus := o.GetFullNodesStatus()
for _, v := range fullStatus {
if !v.Connected {
continue
}
for _, s := range shards {
if dshardorchestrator.ContainsInt(v.Shards, s) {
return ErrShardAlreadyRunning
}
}
}
node := o.FindNodeByID(nodeID)
if node == nil {
return ErrUnknownNode
}
node.StartShards(shards...)
return nil
}
// StopShard will stop the specified shard on whatever node it's running on, or do nothing if it's not running
func (o *Orchestrator) StopShard(shard int) error {
fullStatus := o.GetFullNodesStatus()
for _, v := range fullStatus {
if !v.Connected {
continue
}
if dshardorchestrator.ContainsInt(v.Shards, shard) {
// bingo
node := o.FindNodeByID(v.ID)
if node == nil {
return ErrUnknownNode
}
node.StopShard(shard)
}
}
return nil
}
// MigrateFullNode migrates all the shards on the origin node to the destination node
// optionally also shutting the origin node down at the end
func (o *Orchestrator) MigrateFullNode(fromNode string, toNodeID string, shutdownOldNode bool) error {
nodeFrom := o.FindNodeByID(fromNode)
if nodeFrom == nil {
return ErrUnknownFromNode
}
toNode := o.FindNodeByID(toNodeID)
if toNode == nil {
return ErrUnknownToNode
}
nodeFrom.mu.Lock()
shards := make([]int, len(nodeFrom.runningShards))
copy(shards, nodeFrom.runningShards)
nodeFrom.mu.Unlock()
o.Log(dshardorchestrator.LogInfo, nil, fmt.Sprintf("starting full node migration from %s to %s, n-shards: %d", fromNode, toNodeID, len(shards)))
for _, s := range shards {
err := o.StartShardMigration(toNodeID, s)
if err != nil {
return err
}
// wait for it to be moved before we start the next one
o.WaitForShardMigration(nodeFrom, toNode, s)
// reset here in case something went wrong
toNode.mu.Lock()
toNode.shardMigrationMode = dshardorchestrator.ShardMigrationModeNone
toNode.mu.Unlock()
nodeFrom.mu.Lock()
nodeFrom.shardMigrationMode = dshardorchestrator.ShardMigrationModeNone
nodeFrom.mu.Unlock()
// wait a bit extra to allow for some time ot catch up on events processing
time.Sleep(time.Second)
}
if shutdownOldNode {
return o.ShutdownNode(fromNode)
}
return nil
}
// ShutdownNode shuts down the specified node
func (o *Orchestrator) ShutdownNode(nodeID string) error {
node := o.FindNodeByID(nodeID)
if node == nil {
return ErrUnknownNode
}
node.Shutdown()
return nil
}
// WaitForShardMigration blocks until a shard migration is complete
func (o *Orchestrator) WaitForShardMigration(fromNode *NodeConn, toNode *NodeConn, shardID int) {
// wait for the shard to dissapear on the origin node
for {
time.Sleep(time.Second)
status := fromNode.GetFullStatus()
if !dshardorchestrator.ContainsInt(status.Shards, shardID) || !status.Connected {
// also if we disconnected then just go through all this immeditely
break
}
}
// wait for it to appear on the new node
for {
time.Sleep(time.Millisecond * 100)
statusTo := toNode.GetFullStatus()
statusFrom := fromNode.GetFullStatus()
if dshardorchestrator.ContainsInt(statusTo.Shards, shardID) || !statusFrom.Connected {
// also if we disconnected then just go through all this immeditely
break
}
}
// AND FINALLY just for safe measure, wait for it to not be in the migrating state
for {
time.Sleep(time.Millisecond * 100)
statusTo := toNode.GetFullStatus()
statusFrom := fromNode.GetFullStatus()
if statusTo.MigratingFrom == "" || !statusFrom.Connected {
break
}
}
}
func (o *Orchestrator) findAvailableNode(ignore []*NodeStatus) (string, error) {
numFailed := 0
var lastTimeStartedNode time.Time
for {
// look for a available node
nodes := o.GetFullNodesStatus()
FINDOUTER:
for _, v := range nodes {
if !v.Connected || len(v.Shards) > 0 {
continue
}
for _, ig := range ignore {
if ig.ID == v.ID {
continue FINDOUTER
}
}
if v.MigratingFrom != "" || v.MigratingTo != "" {
continue
}
return v.ID, nil
}
// need to start a new node
// we wait inbetween 10 seconds each launch
// nodes may have been "stolen" by someone in between so we can retry if we fail after 10 seconds
// TODO: reserve nodes
if time.Since(lastTimeStartedNode) < time.Second*60 {
time.Sleep(time.Second)
continue
}
numFailed++
if numFailed > 5 {
return "", errors.New("Failed 5 times")
}
_, err := o.StartNewNode()
if err != nil {
return "", err
}
lastTimeStartedNode = time.Now()
time.Sleep(time.Second)
}
}
// MigrateAllNodesToNewNodes performs a full migration of all nodes
// if returnOnError is set then it will return when one of the nodes fail migrating, otherwise it will just log and continue onto the next
func (o *Orchestrator) MigrateAllNodesToNewNodes(returnOnError bool) error {
o.mu.Lock()
if o.performingFullMigration {
o.mu.Unlock()
return errors.New("Already performing a full migration")
}
o.performingFullMigration = true
o.mu.Unlock()
defer func() {
o.mu.Lock()
o.performingFullMigration = false
o.mu.Unlock()
}()
o.Log(dshardorchestrator.LogInfo, nil, "performing a full node migration on all nodes")
nodes := o.GetFullNodesStatus()
for _, v := range nodes {
if !v.Connected || len(v.Shards) < 1 {
continue
}
targetID, err := o.findAvailableNode(nodes)
if err != nil {
return err
}
err = o.MigrateFullNode(v.ID, targetID, true)
if err != nil {
if returnOnError {
return err
} else {
o.Log(dshardorchestrator.LogError, err, fmt.Sprintf("failed migrating %s to %s", v.ID, targetID))
}
}
// sleep a bit to allow for a buffer for things to settle post migration (maybe some reconnects triggered and such...)
time.Sleep(time.Second * 5)
}
return nil
}
func (o *Orchestrator) getTotalShardCount() (int, error) {
if o.FixedTotalShardCount > 0 {
return o.FixedTotalShardCount, nil
}
return o.ShardCountProvider.GetTotalShardCount()
}
func (o *Orchestrator) isResponsibleForShard(shard int) bool {
if len(o.ResponsibleForShards) < 1 {
return true
}
for _, v := range o.ResponsibleForShards {
if v == shard {
return true
}
}
return false
}
// BlacklistNode blacklists a node from beign designated new shards
func (o *Orchestrator) BlacklistNode(node string) {
o.mu.Lock()
defer o.mu.Unlock()
if o.isNodeBlacklisted(false, node) {
// already blacklisted
return
}
o.blacklistedNodes = append(o.blacklistedNodes, node)
}
func (o *Orchestrator) isNodeBlacklisted(lock bool, node string) bool {
if lock {
o.mu.Lock()
defer o.mu.Unlock()
}
for _, v := range o.blacklistedNodes {
if v == node {
return true
}
}
return false
}