-
Notifications
You must be signed in to change notification settings - Fork 0
/
enginerequests.go
523 lines (445 loc) · 14.7 KB
/
enginerequests.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
package discovery
import (
"context"
"errors"
"fmt"
"sync"
"sync/atomic"
"time"
"github.com/google/uuid"
"github.com/nats-io/nats.go"
"github.com/overmindtech/sdp-go"
log "github.com/sirupsen/logrus"
"github.com/sourcegraph/conc/pool"
"go.opentelemetry.io/otel/attribute"
"go.opentelemetry.io/otel/codes"
"go.opentelemetry.io/otel/trace"
"google.golang.org/protobuf/types/known/durationpb"
"google.golang.org/protobuf/types/known/timestamppb"
)
// AllSourcesFailedError Will be returned when all sources have failed
type AllSourcesFailedError struct {
NumSources int
}
func (e AllSourcesFailedError) Error() string {
return fmt.Sprintf("all sources (%v) failed", e.NumSources)
}
// NewItemSubject Generates a random subject name for returning items e.g.
// return.item._INBOX.712ab421
func NewItemSubject() string {
return fmt.Sprintf("return.item.%v", nats.NewInbox())
}
// NewResponseSubject Generates a random subject name for returning responses
// e.g. return.response._INBOX.978af6de
func NewResponseSubject() string {
return fmt.Sprintf("return.response.%v", nats.NewInbox())
}
// HandleQuery Handles a single query. This includes responses, linking
// etc.
func (e *Engine) HandleQuery(ctx context.Context, query *sdp.Query) {
span := trace.SpanFromContext(ctx)
span.SetName("HandleQuery")
var deadlineOverride bool
// If there is no deadline OR further in the future than MaxRequestTimeout, clamp the deadline to MaxRequestTimeout
maxRequestDeadline := time.Now().Add(e.MaxRequestTimeout)
if query.GetDeadline() == nil || query.GetDeadline().AsTime().After(maxRequestDeadline) {
query.Deadline = timestamppb.New(maxRequestDeadline)
deadlineOverride = true
log.WithContext(ctx).WithField("ovm.deadline", query.GetDeadline().AsTime()).Debug("capping deadline to MaxRequestTimeout")
}
// Add the query timeout to the context stack
ctx, cancel := query.TimeoutContext(ctx)
defer cancel()
numExpandedQueries := len(e.sh.ExpandQuery(query))
// Extract and parse the UUID
u, uuidErr := uuid.FromBytes(query.GetUUID())
span.SetAttributes(
attribute.Int("ovm.discovery.numExpandedQueries", numExpandedQueries),
attribute.String("ovm.sdp.uuid", u.String()),
attribute.String("ovm.sdp.type", query.GetType()),
attribute.String("ovm.sdp.method", query.GetMethod().String()),
attribute.String("ovm.sdp.query", query.GetQuery()),
attribute.String("ovm.sdp.scope", query.GetScope()),
attribute.String("ovm.sdp.deadline", query.GetDeadline().AsTime().String()),
attribute.Bool("ovm.sdp.deadlineOverridden", deadlineOverride),
attribute.Bool("ovm.sdp.queryIgnoreCache", query.GetIgnoreCache()),
)
if query.GetRecursionBehaviour() != nil {
span.SetAttributes(
attribute.Int("ovm.sdp.linkDepth", int(query.GetRecursionBehaviour().GetLinkDepth())),
attribute.Bool("ovm.sdp.followOnlyBlastPropagation", query.GetRecursionBehaviour().GetFollowOnlyBlastPropagation()),
)
}
if numExpandedQueries == 0 {
// If we don't have any relevant sources, exit
return
}
// Respond saying we've got it
responder := sdp.ResponseSender{
ResponseSubject: query.Subject(),
}
var pub sdp.EncodedConnection
if e.IsNATSConnected() {
span.SetAttributes(attribute.Bool("ovm.nats.connected", true))
pub = e.natsConnection
} else {
span.SetAttributes(attribute.Bool("ovm.nats.connected", false))
pub = NilConnection{}
}
ru := uuid.New()
responder.Start(
ctx,
pub,
e.Name,
ru,
)
qt := QueryTracker{
Query: query,
Engine: e,
Context: ctx,
Cancel: cancel,
}
if uuidErr == nil {
e.TrackQuery(u, &qt)
defer e.DeleteTrackedQuery(u)
}
_, _, err := qt.Execute(ctx)
// If all failed then return an error
if err != nil {
if errors.Is(err, context.Canceled) {
responder.CancelWithContext(ctx)
} else {
responder.ErrorWithContext(ctx)
}
span.SetAttributes(
attribute.String("ovm.sdp.errorType", "OTHER"),
attribute.String("ovm.sdp.errorString", err.Error()),
)
} else {
responder.DoneWithContext(ctx)
}
}
// ExecuteQuerySync Executes a Query, waiting for all results, then returns
// them along with the error, rather than passing the results back along channels
func (e *Engine) ExecuteQuerySync(ctx context.Context, q *sdp.Query) ([]*sdp.Item, []*sdp.QueryError, error) {
itemsChan := make(chan *sdp.Item, 100_000)
errsChan := make(chan *sdp.QueryError, 100_000)
items := make([]*sdp.Item, 0)
errs := make([]*sdp.QueryError, 0)
err := e.ExecuteQuery(ctx, q, itemsChan, errsChan)
for i := range itemsChan {
items = append(items, i)
}
for e := range errsChan {
errs = append(errs, e)
}
return items, errs, err
}
var listExecutionPoolCount atomic.Int32
var getExecutionPoolCount atomic.Int32
// ExecuteQuery Executes a single Query and returns the results without any
// linking. Will return an error if all sources fail, or the Query couldn't be
// run.
//
// Items and errors will be sent to the supplied channels as they are found.
// Note that if these channels are not buffered, something will need to be
// receiving the results or this method will never finish. If results are not
// required the channels can be nil
func (e *Engine) ExecuteQuery(ctx context.Context, query *sdp.Query, items chan<- *sdp.Item, errs chan<- *sdp.QueryError) error {
span := trace.SpanFromContext(ctx)
// Make sure we close channels once we're done
if items != nil {
defer close(items)
}
if errs != nil {
defer close(errs)
}
if ctx.Err() != nil {
return ctx.Err()
}
expanded := e.sh.ExpandQuery(query)
span.SetAttributes(
attribute.Int("ovm.source.numExpandedQueries", len(expanded)),
)
if len(expanded) == 0 {
errs <- &sdp.QueryError{
ErrorType: sdp.QueryError_NOSCOPE,
ErrorString: "no matching sources found",
Scope: query.GetScope(),
}
return errors.New("no matching sources found")
}
// These are used to calculate whether all sources have failed or not
var numSources atomic.Int32
var numErrs int
// Since we need to wait for only the processing of this query's executions, we need a separate WaitGroup here
// Overall MaxParallelExecutions evaluation is handled by e.executionPool
wg := sync.WaitGroup{}
for q, sources := range expanded {
wg.Add(1)
// localize values for the closure below
q, sources := q, sources
var p *pool.Pool
if q.GetMethod() == sdp.QueryMethod_LIST {
p = e.listExecutionPool
listExecutionPoolCount.Add(1)
} else {
p = e.getExecutionPool
getExecutionPoolCount.Add(1)
}
// push all queued items through a goroutine to avoid blocking `ExecuteQuery` from progressing
// as `executionPool.Go()` will block once the max parallelism is hit
go func() {
// queue everything into the execution pool
defer LogRecoverToReturn(ctx, "ExecuteQuery outer")
span.SetAttributes(
attribute.Int("ovm.discovery.listExecutionPoolCount", int(listExecutionPoolCount.Load())),
attribute.Int("ovm.discovery.getExecutionPoolCount", int(getExecutionPoolCount.Load())),
)
p.Go(func() {
defer LogRecoverToReturn(ctx, "ExecuteQuery inner")
defer wg.Done()
defer func() {
if q.GetMethod() == sdp.QueryMethod_LIST {
listExecutionPoolCount.Add(-1)
} else {
getExecutionPoolCount.Add(-1)
}
}()
var queryItems []*sdp.Item
var queryErrors []*sdp.QueryError
numSources.Add(1)
// query all sources
queryItems, queryErrors = e.Execute(ctx, q, sources)
for _, i := range queryItems {
// Assign the source query
if i.GetMetadata() != nil {
i.Metadata.SourceQuery = query
}
if items != nil {
items <- i
}
}
for _, e := range queryErrors {
if q != nil {
numErrs++
if errs != nil {
errs <- e
}
}
}
})
}()
}
wg.Wait()
// If all failed then return first error
if numSourcesInt := numSources.Load(); numErrs == int(numSourcesInt) {
return AllSourcesFailedError{
NumSources: int(numSourcesInt),
}
}
return nil
}
// Execute Runs the query against known sources in priority order. If nothing was
// found, returns the first error. This returns a slice if items for
// convenience, but this should always be of length 1 or 0
func (e *Engine) Execute(ctx context.Context, q *sdp.Query, relevantSources []Source) ([]*sdp.Item, []*sdp.QueryError) {
sources := relevantSources
if q.GetMethod() == sdp.QueryMethod_SEARCH {
sources = make([]Source, 0)
// Filter further by searchability
for _, source := range relevantSources {
if searchable, ok := source.(SearchableSource); ok {
sources = append(sources, searchable)
}
}
}
return e.callSources(ctx, q, sources)
}
func (e *Engine) callSources(ctx context.Context, q *sdp.Query, relevantSources []Source) ([]*sdp.Item, []*sdp.QueryError) {
ctx, span := tracer.Start(ctx, "CallSources", trace.WithAttributes(
attribute.String("ovm.source.queryMethod", q.GetMethod().String()),
))
defer span.End()
// Check that our context is okay before doing anything expensive
if ctx.Err() != nil {
span.RecordError(ctx.Err())
return nil, []*sdp.QueryError{
{
UUID: q.GetUUID(),
ErrorType: sdp.QueryError_OTHER,
ErrorString: ctx.Err().Error(),
Scope: q.GetScope(),
ResponderName: e.Name,
ItemType: q.GetType(),
},
}
}
items := make([]*sdp.Item, 0)
errs := make([]*sdp.QueryError, 0)
// We want to avoid having a Get and a List running at the same time, we'd
// rather run the List first, populate the cache, then have the Get just
// grab the value from the cache. To this end we use a GetListMutex to allow
// a List to block all subsequent Get queries until it is done
switch q.GetMethod() {
case sdp.QueryMethod_GET:
e.gfm.GetLock(q.GetScope(), q.GetType())
defer e.gfm.GetUnlock(q.GetScope(), q.GetType())
case sdp.QueryMethod_LIST:
e.gfm.ListLock(q.GetScope(), q.GetType())
defer e.gfm.ListUnlock(q.GetScope(), q.GetType())
case sdp.QueryMethod_SEARCH:
// We don't need to lock for a search since they are independent and
// will only ever have a cache hit if the query is identical
}
span.SetAttributes(
attribute.String("ovm.source.queryType", q.GetType()),
attribute.String("ovm.source.queryScope", q.GetScope()),
)
for _, src := range relevantSources {
if func() bool {
// start querying the source after a cache miss
ctx, span := tracer.Start(ctx, src.Name(), trace.WithAttributes(
attribute.String("ovm.source.method", q.GetMethod().String()),
attribute.String("ovm.source.queryMethod", q.GetMethod().String()),
attribute.String("ovm.source.queryType", q.GetType()),
attribute.String("ovm.source.queryScope", q.GetScope()),
attribute.String("ovm.source.name", src.Name()),
attribute.String("ovm.source.query", q.GetQuery()),
))
defer span.End()
// Ensure that the span is closed when the context is done. This is based on
// the assumption that some sources may not respect the context deadline and
// may run indefinitely. This ensures that we at least get notified about
// it.
go func() {
<-ctx.Done()
if ctx.Err() != nil {
// get a fresh copy of the span to avoid data races
span := trace.SpanFromContext(ctx)
span.RecordError(ctx.Err())
span.SetAttributes(
attribute.Bool("ovm.discover.hang", true),
)
span.End()
}
}()
var resultItems []*sdp.Item
var err error
var sourceDuration time.Duration
start := time.Now()
switch q.GetMethod() {
case sdp.QueryMethod_GET:
var newItem *sdp.Item
newItem, err = src.Get(ctx, q.GetScope(), q.GetQuery(), q.GetIgnoreCache())
if err == nil {
resultItems = []*sdp.Item{newItem}
}
case sdp.QueryMethod_LIST:
resultItems, err = src.List(ctx, q.GetScope(), q.GetIgnoreCache())
case sdp.QueryMethod_SEARCH:
if searchableSrc, ok := src.(SearchableSource); ok {
resultItems, err = searchableSrc.Search(ctx, q.GetScope(), q.GetQuery(), q.GetIgnoreCache())
} else {
err = &sdp.QueryError{
ErrorType: sdp.QueryError_NOTFOUND,
ErrorString: "source is not searchable",
}
}
}
sourceDuration = time.Since(start)
span.SetAttributes(
attribute.Int("ovm.source.numItems", len(resultItems)),
attribute.Bool("ovm.source.cache", false),
attribute.String("ovm.source.duration", sourceDuration.String()),
)
if considerFailed(err) {
span.SetStatus(codes.Error, err.Error())
}
if err != nil {
span.SetAttributes(attribute.String("ovm.source.error", err.Error()))
var sdpErr *sdp.QueryError
if errors.As(err, &sdpErr) {
// Add details if they aren't populated
scope := sdpErr.GetScope()
if scope == "" {
scope = q.GetScope()
}
errs = append(errs, &sdp.QueryError{
UUID: q.GetUUID(),
ErrorType: sdpErr.GetErrorType(),
ErrorString: sdpErr.GetErrorString(),
Scope: scope,
SourceName: src.Name(),
ItemType: src.Type(),
ResponderName: e.Name,
})
} else {
errs = append(errs, &sdp.QueryError{
UUID: q.GetUUID(),
ErrorType: sdp.QueryError_OTHER,
ErrorString: err.Error(),
Scope: q.GetScope(),
SourceName: src.Name(),
ItemType: q.GetType(),
ResponderName: e.Name,
})
}
}
// For each found item, add more details
//
// Use the index here to ensure that we're actually editing the
// right thing
for _, item := range resultItems {
// Handle the case where we are given a nil pointer
if item == nil {
continue
}
// Store metadata
item.Metadata = &sdp.Metadata{
Timestamp: timestamppb.New(time.Now()),
SourceDuration: durationpb.New(sourceDuration),
SourceDurationPerItem: durationpb.New(time.Duration(sourceDuration.Nanoseconds() / int64(len(resultItems)))),
SourceName: src.Name(),
SourceQuery: q,
}
// Mark the item as hidden if the source is a hidden source
if hs, ok := src.(HiddenSource); ok {
item.Metadata.Hidden = hs.Hidden()
}
}
items = append(items, resultItems...)
if q.GetMethod() == sdp.QueryMethod_GET {
// If it's a get, we just return the first thing that works
if len(resultItems) > 0 {
return true
}
}
return false
}() {
// `get` queries only return the first source results
break
}
}
return items, errs
}
// considerFailed Returns whether or not a given error should be considered as a
// failure or not. The only error that isn't consider a failure is a
// *sdp.QueryError with a Type of NOTFOUND, this means that it was queried
// successfully but it simply doesn't exist
func considerFailed(err error) bool {
if err == nil {
return false
} else {
var sdpErr *sdp.QueryError
if errors.As(err, &sdpErr) {
if sdpErr.GetErrorType() == sdp.QueryError_NOTFOUND {
return false
} else {
return true
}
} else {
return true
}
}
}