forked from purpleidea/mgmt
-
Notifications
You must be signed in to change notification settings - Fork 0
/
main.go
845 lines (737 loc) · 26.4 KB
/
main.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
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
// Mgmt
// Copyright (C) 2013-2018+ James Shubin and the project contributors
// Written by James Shubin <james@shubin.ca> and the project contributors
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
package lib
import (
"fmt"
"io/ioutil"
"log"
"os"
"path"
"sync"
"time"
"github.com/purpleidea/mgmt/converger"
"github.com/purpleidea/mgmt/etcd"
"github.com/purpleidea/mgmt/gapi"
"github.com/purpleidea/mgmt/pgp"
"github.com/purpleidea/mgmt/pgraph"
"github.com/purpleidea/mgmt/prometheus"
"github.com/purpleidea/mgmt/recwatch"
"github.com/purpleidea/mgmt/remote"
"github.com/purpleidea/mgmt/resources"
"github.com/purpleidea/mgmt/util"
etcdtypes "github.com/coreos/etcd/pkg/types"
multierr "github.com/hashicorp/go-multierror"
errwrap "github.com/pkg/errors"
)
// Flags are some constant flags which are used throughout the program.
type Flags struct {
Debug bool // add additional log messages
Trace bool // add execution flow log messages
Verbose bool // add extra log message output
}
// Main is the main struct for running the mgmt logic.
type Main struct {
Program string // the name of this program, usually set at compile time
Version string // the version of this program, usually set at compile time
Flags Flags // static global flags that are set at compile time
Hostname *string // hostname to use; nil if undefined
Prefix *string // prefix passed in; nil if undefined
TmpPrefix bool // request a pseudo-random, temporary prefix to be used
AllowTmpPrefix bool // allow creation of a new temporary prefix if main prefix is unavailable
Deploy *gapi.Deploy // deploy object including GAPI for static deploys
DeployFs resources.Fs // used for static deploys
Remotes []string // list of remote graph definitions to run
NoWatch bool // do not change graph under any circumstances
NoConfigWatch bool // do not update graph due to config changes
NoStreamWatch bool // do not update graph due to stream changes
Noop bool // globally force all resources into no-op mode
Sema int // add a semaphore with this lock count to each resource
Graphviz string // output file for graphviz data
GraphvizFilter string // graphviz filter to use
ConvergedTimeout int // exit after approximately this many seconds in a converged state; -1 to disable
MaxRuntime uint // exit after a maximum of approximately this many seconds
Seeds []string // default etc client endpoint
ClientURLs []string // list of URLs to listen on for client traffic
ServerURLs []string // list of URLs to listen on for server (peer) traffic
AdvertiseClientURLs []string // list of URLs to advertise for client traffic
AdvertiseServerURLs []string // list of URLs to advertise for server (peer) traffic
IdealClusterSize int // ideal number of server peers in cluster; only read by initial server
NoServer bool // do not let other servers peer with me
CConns uint16 // number of maximum concurrent remote ssh connections to run, 0 for unlimited
AllowInteractive bool // allow interactive prompting, such as for remote passwords
SSHPrivIDRsa string // default path to ssh key file, set empty to never touch
NoCaching bool // don't allow remote caching of remote execution binary
Depth uint16 // depth in remote hierarchy; for internal use only
seeds etcdtypes.URLs // processed seeds value
clientURLs etcdtypes.URLs // processed client urls value
serverURLs etcdtypes.URLs // processed server urls value
advertiseClientURLs etcdtypes.URLs // processed advertise client urls value
advertiseServerURLs etcdtypes.URLs // processed advertise server urls value
idealClusterSize uint16 // processed ideal cluster size value
NoPgp bool // disallow pgp functionality
PgpKeyPath *string // import a pre-made key pair
PgpIdentity *string
pgpKeys *pgp.PGP // agent key pair
Prometheus bool // enable prometheus metrics
PrometheusListen string // prometheus instance bind specification
exit chan error // exit signal
}
// Init initializes the main struct after it performs some validation.
func (obj *Main) Init() error {
if obj.Program == "" || obj.Version == "" {
return fmt.Errorf("you must set the Program and Version strings")
}
if obj.Prefix != nil && obj.TmpPrefix {
return fmt.Errorf("choosing a prefix and the request for a tmp prefix is illogical")
}
// if we've turned off watching, then be explicit and disable them all!
// if all the watches are disabled, then it's equivalent to no watching
if obj.NoWatch {
obj.NoConfigWatch = true
obj.NoStreamWatch = true
} else if obj.NoConfigWatch && obj.NoStreamWatch {
obj.NoWatch = true
}
obj.idealClusterSize = uint16(obj.IdealClusterSize)
if obj.IdealClusterSize < 0 { // value is undefined, set to the default
obj.idealClusterSize = etcd.DefaultIdealClusterSize
}
if obj.idealClusterSize < 1 {
return fmt.Errorf("the IdealClusterSize should be at least one")
}
if obj.NoServer && len(obj.Remotes) > 0 {
// TODO: in this case, we won't be able to tunnel stuff back to
// here, so if we're okay with every remote graph running in an
// isolated mode, then this is okay. Improve on this if there's
// someone who really wants to be able to do this.
return fmt.Errorf("the Server is required when using Remotes")
}
if obj.CConns < 0 {
return fmt.Errorf("the CConns value should be at least zero")
}
if obj.ConvergedTimeout >= 0 && obj.CConns > 0 && len(obj.Remotes) > int(obj.CConns) {
return fmt.Errorf("you can't converge if you have more remotes than available connections")
}
if obj.Depth < 0 { // user should not be using this argument manually
return fmt.Errorf("negative values for Depth are not permitted")
}
// transform the url list inputs into etcd typed lists
var err error
obj.seeds, err = etcdtypes.NewURLs(
util.FlattenListWithSplit(obj.Seeds, []string{",", ";", " "}),
)
if err != nil && len(obj.Seeds) > 0 {
return errwrap.Wrapf(err, "the Seeds didn't parse correctly")
}
obj.clientURLs, err = etcdtypes.NewURLs(
util.FlattenListWithSplit(obj.ClientURLs, []string{",", ";", " "}),
)
if err != nil && len(obj.ClientURLs) > 0 {
return errwrap.Wrapf(err, "the ClientURLs didn't parse correctly")
}
obj.serverURLs, err = etcdtypes.NewURLs(
util.FlattenListWithSplit(obj.ServerURLs, []string{",", ";", " "}),
)
if err != nil && len(obj.ServerURLs) > 0 {
return errwrap.Wrapf(err, "the ServerURLs didn't parse correctly")
}
obj.advertiseClientURLs, err = etcdtypes.NewURLs(
util.FlattenListWithSplit(obj.AdvertiseClientURLs, []string{",", ";", " "}),
)
if err != nil && len(obj.AdvertiseClientURLs) > 0 {
return errwrap.Wrapf(err, "the AdvertiseClientURLs didn't parse correctly")
}
obj.advertiseServerURLs, err = etcdtypes.NewURLs(
util.FlattenListWithSplit(obj.AdvertiseServerURLs, []string{",", ";", " "}),
)
if err != nil && len(obj.AdvertiseServerURLs) > 0 {
return errwrap.Wrapf(err, "the AdvertiseServerURLs didn't parse correctly")
}
obj.exit = make(chan error)
return nil
}
// Exit causes a safe shutdown. This is often attached to the ^C signal handler.
func (obj *Main) Exit(err error) {
obj.exit <- err // trigger an exit!
}
// Run is the main execution entrypoint to run mgmt.
func (obj *Main) Run() error {
hello(obj.Program, obj.Version, obj.Flags) // say hello!
hostname, err := os.Hostname() // a sensible default
// allow passing in the hostname, instead of using the system setting
if h := obj.Hostname; h != nil && *h != "" { // override by cli
hostname = *h
} else if err != nil {
return errwrap.Wrapf(err, "can't get default hostname")
}
if hostname == "" { // safety check
return fmt.Errorf("hostname cannot be empty")
}
var prefix = fmt.Sprintf("/var/lib/%s/", obj.Program) // default prefix
if p := obj.Prefix; p != nil {
prefix = *p
}
// make sure the working directory prefix exists
if obj.TmpPrefix || os.MkdirAll(prefix, 0770) != nil {
if obj.TmpPrefix || obj.AllowTmpPrefix {
var err error
if prefix, err = ioutil.TempDir("", obj.Program+"-"+hostname+"-"); err != nil {
return fmt.Errorf("can't create temporary prefix")
}
log.Println("Main: Warning: Working prefix directory is temporary!")
} else {
return fmt.Errorf("can't create prefix")
}
}
log.Printf("Main: Working prefix is: %s", prefix)
pgraphPrefix := fmt.Sprintf("%s/", path.Join(prefix, "pgraph")) // pgraph namespace
if err := os.MkdirAll(pgraphPrefix, 0770); err != nil {
return errwrap.Wrapf(err, "can't create pgraph prefix")
}
var prom *prometheus.Prometheus
if obj.Prometheus {
prom = &prometheus.Prometheus{
Listen: obj.PrometheusListen,
}
if err := prom.Init(); err != nil {
return errwrap.Wrapf(err, "can't create initiate Prometheus instance")
}
log.Printf("Main: Prometheus: Starting instance on %s", prom.Listen)
if err := prom.Start(); err != nil {
return errwrap.Wrapf(err, "can't start initiate Prometheus instance")
}
if err := prom.InitKindMetrics(resources.RegisteredResourcesNames()); err != nil {
return errwrap.Wrapf(err, "can't initialize kind-specific prometheus metrics")
}
}
if !obj.NoPgp {
pgpPrefix := fmt.Sprintf("%s/", path.Join(prefix, "pgp"))
if err := os.MkdirAll(pgpPrefix, 0770); err != nil {
return errwrap.Wrapf(err, "can't create pgp prefix")
}
pgpKeyringPath := path.Join(pgpPrefix, pgp.DefaultKeyringFile) // default path
if p := obj.PgpKeyPath; p != nil {
pgpKeyringPath = *p
}
var err error
if obj.pgpKeys, err = pgp.Import(pgpKeyringPath); err != nil && !os.IsNotExist(err) {
return errwrap.Wrapf(err, "can't import pgp key")
}
if obj.pgpKeys == nil {
identity := fmt.Sprintf("%s <%s> %s", obj.Program, "root@"+hostname, "generated by "+obj.Program)
if p := obj.PgpIdentity; p != nil {
identity = *p
}
name, comment, email, err := pgp.ParseIdentity(identity)
if err != nil {
return errwrap.Wrapf(err, "can't parse user string")
}
// TODO: Make hash configurable
if obj.pgpKeys, err = pgp.Generate(name, comment, email, nil); err != nil {
return errwrap.Wrapf(err, "can't create pgp key")
}
if err := obj.pgpKeys.SaveKey(pgpKeyringPath); err != nil {
return errwrap.Wrapf(err, "can't save pgp key")
}
}
// TODO: Import admin key
}
oldGraph := &pgraph.Graph{}
graph := &resources.MGraph{}
// pass in the information we need
graph.Debug = obj.Flags.Debug
graph.Init()
// exit after `max-runtime` seconds for no reason at all...
if i := obj.MaxRuntime; i > 0 {
go func() {
time.Sleep(time.Duration(i) * time.Second)
obj.Exit(nil)
}()
}
// setup converger
converger := converger.NewConverger(
obj.ConvergedTimeout,
nil, // stateFn gets added in by EmbdEtcd
)
go converger.Loop(true) // main loop for converger, true to start paused
// TODO: will this shut things down prematurely?
converger.Start() // better start this anyways...
// embedded etcd
if len(obj.seeds) == 0 {
log.Printf("Main: Seeds: No seeds specified!")
} else {
log.Printf("Main: Seeds(%d): %v", len(obj.seeds), obj.seeds)
}
EmbdEtcd := etcd.NewEmbdEtcd(
hostname,
obj.seeds,
obj.clientURLs,
obj.serverURLs,
obj.advertiseClientURLs,
obj.advertiseServerURLs,
obj.NoServer,
obj.idealClusterSize,
etcd.Flags{
Debug: obj.Flags.Debug,
Trace: obj.Flags.Trace,
Verbose: obj.Flags.Verbose,
},
prefix,
converger,
)
if EmbdEtcd == nil {
// TODO: verify EmbdEtcd is not nil below...
obj.Exit(fmt.Errorf("Main: Etcd: Creation failed"))
} else if err := EmbdEtcd.Startup(); err != nil { // startup (returns when etcd main loop is running)
obj.Exit(fmt.Errorf("Main: Etcd: Startup failed: %v", err))
}
// wait for etcd server to be ready before continuing...
// XXX: this is wrong if we're not going to be a server! we'll block!!!
// select {
// case <-EmbdEtcd.ServerReady():
// log.Printf("Main: Etcd: Server: Ready!")
// // pass
// case <-time.After(((etcd.MaxStartServerTimeout * etcd.MaxStartServerRetries) + 1) * time.Second):
// obj.Exit(fmt.Errorf("Main: Etcd: Startup timeout"))
// }
time.Sleep(1 * time.Second) // XXX: temporary workaround
convergerStateFn := func(b bool) error {
// exit if we are using the converged timeout and we are the
// root node. otherwise, if we are a child node in a remote
// execution hierarchy, we should only notify our converged
// state and wait for the parent to trigger the exit.
if t := obj.ConvergedTimeout; obj.Depth == 0 && t >= 0 {
if b {
log.Printf("Main: Converged for %d seconds, exiting!", t)
obj.Exit(nil) // trigger an exit!
}
return nil
}
// send our individual state into etcd for others to see
return etcd.SetHostnameConverged(EmbdEtcd, hostname, b) // TODO: what should happen on error?
}
if EmbdEtcd != nil {
converger.SetStateFn(convergerStateFn)
}
// implementation of the World API (alternates can be substituted in)
world := &etcd.World{
Hostname: hostname,
EmbdEtcd: EmbdEtcd,
MetadataPrefix: MetadataPrefix,
StoragePrefix: StoragePrefix,
StandaloneFs: obj.DeployFs, // used for static deploys
Debug: obj.Flags.Debug,
Logf: func(format string, v ...interface{}) {
log.Printf("world: etcd: "+format, v...)
},
}
graph.Data = &resources.ResData{
Hostname: hostname,
Converger: converger,
Prometheus: prom,
World: world,
Prefix: pgraphPrefix,
Debug: obj.Flags.Debug,
Logf: func(format string, v ...interface{}) {
log.Printf("resources: "+format, v...)
},
}
exitchan := make(chan struct{}) // exit on close
wg := &sync.WaitGroup{} // waitgroup for inner loop (go routine)
deployChan := make(chan *gapi.Deploy)
var gapiImpl gapi.GAPI // active GAPI implementation
gapiImpl = nil // starts off missing
var gapiChan chan gapi.Next // stream events contain some instructions!
gapiChan = nil // starts off blocked
wg.Add(1)
go func() {
defer wg.Done()
first := true // first loop or not
var mainDeploy *gapi.Deploy
for {
log.Println("Main: Waiting...")
// The GAPI should always kick off an event on Next() at
// startup when (and if) it indeed has a graph to share!
fastPause := false
select {
case deploy, ok := <-deployChan:
if !ok { // channel closed
if obj.Flags.Debug {
log.Printf("Main: Deploy: exited")
}
deployChan = nil // disable it
if gapiImpl != nil { // currently running...
gapiChan = nil
if err := gapiImpl.Close(); err != nil {
err = errwrap.Wrapf(err, "the GAPI closed poorly")
log.Printf("Main: Deploy: GAPI: Final close failed: %+v", err)
}
}
return // this is the only place we exit
}
if deploy == nil {
log.Printf("Main: Deploy: Received empty deploy")
continue
}
mainDeploy = deploy // save this one
gapiObj := mainDeploy.GAPI
if gapiObj == nil {
log.Printf("Main: Deploy: Received empty GAPI")
continue
}
if gapiImpl != nil { // currently running...
gapiChan = nil
if err := gapiImpl.Close(); err != nil {
err = errwrap.Wrapf(err, "the GAPI closed poorly")
log.Printf("Main: Deploy: GAPI: Close failed: %+v", err)
}
}
gapiImpl = gapiObj // copy it to active
data := gapi.Data{
Program: obj.Program,
Hostname: hostname,
World: world,
Noop: mainDeploy.Noop,
// FIXME: should the below flags come from the deploy struct?
//NoWatch: obj.NoWatch,
NoConfigWatch: obj.NoConfigWatch,
NoStreamWatch: obj.NoStreamWatch,
Debug: obj.Flags.Debug,
Logf: func(format string, v ...interface{}) {
log.Printf("gapi: "+format, v...)
},
}
if obj.Flags.Debug {
log.Printf("Main: GAPI: Init...")
}
if err := gapiImpl.Init(data); err != nil {
log.Printf("Main: GAPI: Init failed: %+v", err)
// TODO: consider running previous GAPI?
} else {
if obj.Flags.Debug {
log.Printf("Main: GAPI: Next...")
}
// this must generate at least one event for it to work
gapiChan = gapiImpl.Next() // stream of graph switch events!
}
continue
case next, ok := <-gapiChan:
if !ok { // channel closed
if obj.Flags.Debug {
log.Printf("Main: GAPI exited")
}
gapiChan = nil // disable it
continue
}
// if we've been asked to exit...
// TODO: do we want to block exits and wait?
// TODO: we might want to wait for the next GAPI
if next.Exit {
obj.Exit(next.Err) // trigger exit
continue // wait for exitchan
}
// the gapi lets us send an error to the channel
// this means there was a failure, but not fatal
if err := next.Err; err != nil {
log.Printf("Main: Error with graph stream: %v", err)
continue // wait for another event
}
// everything else passes through to cause a compile!
fastPause = next.Fast // should we pause fast?
//case <-exitchan: // we only exit on deployChan close!
// return
}
if gapiImpl == nil { // TODO: can this ever happen anymore?
log.Printf("Main: GAPI is empty!")
continue
}
if first {
converger.Pause() // it's already started!
}
// we need the vertices to be paused to work on them, so
// run graph vertex LOCK...
if !first { // TODO: we can flatten this check out I think
converger.Pause() // FIXME: add sync wait?
graph.Pause(fastPause) // sync
//graph.UnGroup() // FIXME: implement me if needed!
}
// make the graph from yaml, lib, puppet->yaml, or dsl!
newGraph, err := gapiImpl.Graph() // generate graph!
if err != nil {
log.Printf("Main: Error creating new graph: %v", err)
// unpause!
if !first {
graph.Start(first) // sync
converger.Start() // after Start()
}
continue
}
if obj.Flags.Debug {
log.Printf("Main: New Graph: %v", newGraph)
}
// this edits the paused vertices, but it is safe to do
// so even if we don't use this new graph, since those
// value should be the same for existing vertices...
for _, v := range newGraph.Vertices() {
m := resources.VtoR(v).Meta()
// apply the global noop parameter if requested
if mainDeploy.Noop {
m.Noop = mainDeploy.Noop
}
// append the semaphore to each resource
if mainDeploy.Sema > 0 { // NOTE: size == 0 would block
// a semaphore with an empty id is valid
m.Sema = append(m.Sema, fmt.Sprintf(":%d", mainDeploy.Sema))
}
}
// We don't have to "UnGroup()" to compare, since we
// save the old graph to use when we compare.
// TODO: Does this hurt performance or graph changes ?
log.Printf("Main: GraphSync...")
vertexCmpFn := func(v1, v2 pgraph.Vertex) (bool, error) {
return resources.VtoR(v1).Compare(resources.VtoR(v2)), nil
}
vertexAddFn := func(v pgraph.Vertex) error {
err := resources.VtoR(v).Validate()
return errwrap.Wrapf(err, "could not Validate() resource")
}
vertexRemoveFn := func(v pgraph.Vertex) error {
// wait for exit before starting new graph!
resources.VtoR(v).Exit() // sync
return nil
}
edgeCmpFn := func(e1, e2 pgraph.Edge) (bool, error) {
edge1 := e1.(*resources.Edge) // panic if wrong
edge2 := e2.(*resources.Edge) // panic if wrong
return edge1.Compare(edge2), nil
}
// on success, this updates the receiver graph...
if err := oldGraph.GraphSync(newGraph, vertexCmpFn, vertexAddFn, vertexRemoveFn, edgeCmpFn); err != nil {
log.Printf("Main: Error running graph sync: %v", err)
// unpause!
if !first {
graph.Start(first) // sync
converger.Start() // after Start()
}
continue
}
//savedGraph := oldGraph.Copy() // save a copy for errors
// TODO: should we call each Res.Setup() here instead?
// add autoedges; modifies the graph only if no error
if err := resources.AutoEdges(oldGraph); err != nil {
log.Printf("Main: Error running auto edges: %v", err)
// unpause!
if !first {
graph.Start(first) // sync
converger.Start() // after Start()
}
continue
}
// at this point, any time we error after a destructive
// modification of the graph we need to restore the old
// graph that was previously running, eg:
//
// oldGraph = savedGraph.Copy()
//
// which we are (luckily) able to avoid testing for now
resources.AutoGroup(oldGraph, &resources.NonReachabilityGrouper{}) // run autogroup; modifies the graph
// TODO: do we want to do a transitive reduction?
// FIXME: run a type checker that verifies all the send->recv relationships
graph.Update(oldGraph) // copy in structure of new graph
// Call this here because at this point the graph does
// not know anything about the prometheus instance.
if err := prom.UpdatePgraphStartTime(); err != nil {
log.Printf("Main: Prometheus.UpdatePgraphStartTime() errored: %v", err)
}
// Start() needs to be synchronous or wait,
// because if half of the nodes are started and
// some are not ready yet and the EtcdWatch
// loops, we'll cause Pause() before we
// even got going, thus causing nil pointer errors
graph.Start(first) // sync
converger.Start() // after Start()
log.Printf("Main: Graph: %v", graph) // show graph
if obj.Graphviz != "" {
filter := obj.GraphvizFilter
if filter == "" {
filter = "dot" // directed graph default
}
if err := graph.ExecGraphviz(filter, obj.Graphviz, hostname); err != nil {
log.Printf("Main: Graphviz: %v", err)
} else {
log.Printf("Main: Graphviz: Successfully generated graph!")
}
}
first = false
}
}()
configWatcher := recwatch.NewConfigWatcher()
configWatcher.Flags = recwatch.Flags{Debug: obj.Flags.Debug}
events := configWatcher.Events()
if !obj.NoWatch { // FIXME: fit this into a clean GAPI?
configWatcher.Add(obj.Remotes...) // add all the files...
} else {
events = nil // signal that no-watch is true
}
go func() {
select {
case err := <-configWatcher.Error():
obj.Exit(err) // trigger an exit!
case <-exitchan:
return
}
}()
// initialize the add watcher, which calls the f callback on map changes
convergerCb := func(f func(map[string]bool) error) (func(), error) {
return etcd.AddHostnameConvergedWatcher(EmbdEtcd, f)
}
// build remotes struct for remote ssh
remotes := remote.NewRemotes(
EmbdEtcd.LocalhostClientURLs().StringSlice(),
[]string{etcd.DefaultClientURL},
obj.Noop,
obj.Remotes, // list of files
events, // watch for file changes
obj.CConns,
obj.AllowInteractive,
obj.SSHPrivIDRsa,
!obj.NoCaching,
obj.Depth,
prefix,
converger,
convergerCb,
remote.Flags{
Program: obj.Program,
Debug: obj.Flags.Debug,
},
)
// TODO: is there any benefit to running the remotes above in the loop?
// wait for etcd to be running before we remote in, which we do above!
go remotes.Run()
// wait for remotes to be ready before continuing...
select {
case <-remotes.Ready():
log.Printf("Main: Remotes: Run: Ready!")
// pass
//case <-time.After( ? * time.Second):
// obj.Exit(fmt.Errorf("Main: Remotes: Run timeout"))
}
if obj.Deploy != nil {
deploy := obj.Deploy
// redundant
deploy.Noop = obj.Noop
deploy.Sema = obj.Sema
select {
case deployChan <- deploy:
// send
case <-exitchan:
// pass
}
// don't inline this, because when we close the deployChan it's
// the signal to tell the engine to actually shutdown...
go func() {
defer close(deployChan) // no more are coming ever!
select { // wait until we're ready to shutdown
case <-exitchan:
return
}
}()
} else {
// etcd based deploy
go func() {
defer close(deployChan)
startChan := make(chan struct{}) // start signal
close(startChan) // kick it off!
for {
select {
case <-startChan: // kick the loop once at start
startChan = nil // disable
case err, ok := <-etcd.WatchDeploy(EmbdEtcd):
if !ok {
obj.Exit(nil) // regular shutdown
return
}
if err != nil {
// TODO: it broke, can we restart?
obj.Exit(fmt.Errorf("Main: Deploy: Watch error"))
return
}
case <-exitchan:
return
}
if obj.Flags.Debug {
log.Printf("Main: Deploy: Got activity")
}
str, err := etcd.GetDeploy(EmbdEtcd, 0) // 0 means get the latest one
if err != nil {
log.Printf("Main: Deploy: Error getting deploy %+v", err)
continue
}
if str == "" { // no available deploys exist yet
continue
}
// decode the deploy (incl. GAPI) and send it!
deploy, err := gapi.NewDeployFromB64(str)
if err != nil {
log.Printf("Main: Deploy: Error decoding deploy %+v", err)
continue
}
select {
case deployChan <- deploy:
// send
if obj.Flags.Debug {
log.Printf("Main: Deploy: Sending new GAPI")
}
case <-exitchan:
return
}
}
}()
}
log.Println("Main: Running...")
reterr := <-obj.exit // wait for exit signal
log.Println("Main: Destroy...")
configWatcher.Close() // stop sending file changes to remotes
if err := remotes.Exit(); err != nil { // tell all the remote connections to shutdown; waits!
err = errwrap.Wrapf(err, "the Remote exited poorly")
reterr = multierr.Append(reterr, err) // list of errors
}
// tell inner main loop to exit
close(exitchan)
wg.Wait()
graph.Exit() // tells all the children to exit, and waits for them to do so
// cleanup etcd main loop last so it can process everything first
if err := EmbdEtcd.Destroy(); err != nil { // shutdown and cleanup etcd
err = errwrap.Wrapf(err, "embedded Etcd exited poorly")
reterr = multierr.Append(reterr, err) // list of errors
}
if obj.Prometheus {
log.Printf("Main: Prometheus: Stopping instance")
if err := prom.Stop(); err != nil {
err = errwrap.Wrapf(err, "the Prometheus instance exited poorly")
reterr = multierr.Append(reterr, err)
}
}
if obj.Flags.Debug {
log.Printf("Main: Graph: %v", graph)
}
if reterr != nil {
log.Printf("Main: Error: %v", reterr)
}
log.Println("Goodbye!")
return reterr
}