forked from couchbase/gocbcore
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathretry.go
310 lines (244 loc) · 12.8 KB
/
retry.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
package gocbcore
import (
"encoding/json"
"math"
"time"
"github.com/couchbase/gocbcore/v9/memd"
)
// RetryRequest is a request that can possibly be retried.
type RetryRequest interface {
RetryAttempts() uint32
Identifier() string
Idempotent() bool
RetryReasons() []RetryReason
retryStrategy() RetryStrategy
recordRetryAttempt(reason RetryReason)
}
// RetryReason represents the reason for an operation possibly being retried.
type RetryReason interface {
AllowsNonIdempotentRetry() bool
AlwaysRetry() bool
Description() string
}
type retryReason struct {
allowsNonIdempotentRetry bool
alwaysRetry bool
description string
}
func (rr retryReason) AllowsNonIdempotentRetry() bool {
return rr.allowsNonIdempotentRetry
}
func (rr retryReason) AlwaysRetry() bool {
return rr.alwaysRetry
}
func (rr retryReason) Description() string {
return rr.description
}
func (rr retryReason) String() string {
return rr.description
}
func (rr retryReason) MarshalJSON() ([]byte, error) {
return json.Marshal(rr.description)
}
var (
// UnknownRetryReason indicates that the operation failed for an unknown reason.
UnknownRetryReason = retryReason{allowsNonIdempotentRetry: false, alwaysRetry: false, description: "UNKNOWN"}
// SocketNotAvailableRetryReason indicates that the operation failed because the underlying socket was not available.
SocketNotAvailableRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "SOCKET_NOT_AVAILABLE"}
// ServiceNotAvailableRetryReason indicates that the operation failed because the requested service was not available.
ServiceNotAvailableRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "SERVICE_NOT_AVAILABLE"}
// NodeNotAvailableRetryReason indicates that the operation failed because the requested node was not available.
NodeNotAvailableRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "NODE_NOT_AVAILABLE"}
// KVNotMyVBucketRetryReason indicates that the operation failed because it was sent to the wrong node for the vbucket.
KVNotMyVBucketRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: true, description: "KV_NOT_MY_VBUCKET"}
// KVCollectionOutdatedRetryReason indicates that the operation failed because the collection ID on the request is outdated.
KVCollectionOutdatedRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: true, description: "KV_COLLECTION_OUTDATED"}
// KVErrMapRetryReason indicates that the operation failed for an unsupported reason but the KV error map indicated
// that the operation can be retried.
KVErrMapRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "KV_ERROR_MAP_RETRY_INDICATED"}
// KVLockedRetryReason indicates that the operation failed because the document was locked.
KVLockedRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "KV_LOCKED"}
// KVTemporaryFailureRetryReason indicates that the operation failed because of a temporary failure.
KVTemporaryFailureRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "KV_TEMPORARY_FAILURE"}
// KVSyncWriteInProgressRetryReason indicates that the operation failed because a sync write is in progress.
KVSyncWriteInProgressRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "KV_SYNC_WRITE_IN_PROGRESS"}
// KVSyncWriteRecommitInProgressRetryReason indicates that the operation failed because a sync write recommit is in progress.
KVSyncWriteRecommitInProgressRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "KV_SYNC_WRITE_RE_COMMIT_IN_PROGRESS"}
// ServiceResponseCodeIndicatedRetryReason indicates that the operation failed and the service responded stating that
// the request should be retried.
ServiceResponseCodeIndicatedRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "SERVICE_RESPONSE_CODE_INDICATED"}
// SocketCloseInFlightRetryReason indicates that the operation failed because the socket was closed whilst the operation
// was in flight.
SocketCloseInFlightRetryReason = retryReason{allowsNonIdempotentRetry: false, alwaysRetry: false, description: "SOCKET_CLOSED_WHILE_IN_FLIGHT"}
// PipelineOverloadedRetryReason indicates that the operation failed because the pipeline queue was full.
PipelineOverloadedRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: true, description: "PIPELINE_OVERLOADED"}
// CircuitBreakerOpenRetryReason indicates that the operation failed because the circuit breaker for the underlying socket was open.
CircuitBreakerOpenRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "CIRCUIT_BREAKER_OPEN"}
// QueryIndexNotFoundRetryReason indicates that the operation failed to to a missing query index
QueryIndexNotFoundRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "QUERY_INDEX_NOT_FOUND"}
// QueryPreparedStatementFailureRetryReason indicates that the operation failed due to a prepared statement failure
QueryPreparedStatementFailureRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "QUERY_PREPARED_STATEMENT_FAILURE"}
// AnalyticsTemporaryFailureRetryReason indicates that an analytics operation failed due to a temporary failure
AnalyticsTemporaryFailureRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "ANALYTICS_TEMPORARY_FAILURE"}
// SearchTooManyRequestsRetryReason indicates that a search operation failed due to too many requests
SearchTooManyRequestsRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "SEARCH_TOO_MANY_REQUESTS"}
// NotReadyRetryReason indicates that the WaitUntilReady operation is not ready
NotReadyRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: true, description: "NOT_READY"}
// NoPipelineSnapshotRetryReason indicates that there was no pipeline snapshot available
NoPipelineSnapshotRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "NO_PIPELINE_SNAPSHOT"}
// BucketNotReadyReason indicates that the user has priviledges to access the bucket but the bucket doesn't exist
// or is in warm up.
// Uncommitted: This API may change in the future.
BucketNotReadyReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "BUCKET_NOT_FOUND"}
// ConnectionErrorRetryReason indicates that there were errors reported by underlying connections
ConnectionErrorRetryReason = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: false, description: "CONNECTION_ERROR"}
// MemdWriteFailure indicates that the operation failed because the write failed on the connection.
MemdWriteFailure = retryReason{allowsNonIdempotentRetry: true, alwaysRetry: true, description: "MEMD_WRITE_FAILURE"}
)
// MaybeRetryRequest will possibly retry a request according to the strategy belonging to the request.
// It will use the reason to determine whether or not the failure reason is one that can be retried.
func (agent *Agent) MaybeRetryRequest(req RetryRequest, reason RetryReason) (bool, time.Time) {
return retryOrchMaybeRetry(req, reason)
}
// RetryAction is used by a RetryStrategy to calculate the duration to wait before retrying an operation.
// Returning a value of 0 indicates to not retry.
type RetryAction interface {
Duration() time.Duration
}
// NoRetryRetryAction represents an action that indicates to not retry.
type NoRetryRetryAction struct {
}
// Duration is the length of time to wait before retrying an operation.
func (ra *NoRetryRetryAction) Duration() time.Duration {
return 0
}
// WithDurationRetryAction represents an action that indicates to retry with a given duration.
type WithDurationRetryAction struct {
WithDuration time.Duration
}
// Duration is the length of time to wait before retrying an operation.
func (ra *WithDurationRetryAction) Duration() time.Duration {
return ra.WithDuration
}
// RetryStrategy is to determine if an operation should be retried, and if so how long to wait before retrying.
type RetryStrategy interface {
RetryAfter(req RetryRequest, reason RetryReason) RetryAction
}
// retryOrchMaybeRetry will possibly retry an operation according to the strategy belonging to the request.
// It will use the reason to determine whether or not the failure reason is one that can be retried.
func retryOrchMaybeRetry(req RetryRequest, reason RetryReason) (bool, time.Time) {
if reason.AlwaysRetry() {
duration := ControlledBackoff(req.RetryAttempts())
logDebugf("Will retry request. Backoff=%s, OperationID=%s. Reason=%s", duration, req.Identifier(), reason)
req.recordRetryAttempt(reason)
return true, time.Now().Add(duration)
}
retryStrategy := req.retryStrategy()
if retryStrategy == nil {
return false, time.Time{}
}
action := retryStrategy.RetryAfter(req, reason)
if action == nil {
logDebugf("Won't retry request. OperationID=%s. Reason=%s", req.Identifier(), reason)
return false, time.Time{}
}
duration := action.Duration()
if duration == 0 {
logDebugf("Won't retry request. OperationID=%s. Reason=%s", req.Identifier(), reason)
return false, time.Time{}
}
logDebugf("Will retry request. Backoff=%s, OperationID=%s. Reason=%s", duration, req.Identifier(), reason)
req.recordRetryAttempt(reason)
return true, time.Now().Add(duration)
}
// failFastRetryStrategy represents a strategy that will never retry.
type failFastRetryStrategy struct {
}
// newFailFastRetryStrategy returns a new FailFastRetryStrategy.
func newFailFastRetryStrategy() *failFastRetryStrategy {
return &failFastRetryStrategy{}
}
// RetryAfter calculates and returns a RetryAction describing how long to wait before retrying an operation.
func (rs *failFastRetryStrategy) RetryAfter(req RetryRequest, reason RetryReason) RetryAction {
return &NoRetryRetryAction{}
}
// BackoffCalculator is used by retry strategies to calculate backoff durations.
type BackoffCalculator func(retryAttempts uint32) time.Duration
// BestEffortRetryStrategy represents a strategy that will keep retrying until it succeeds (or the caller times out
// the request).
type BestEffortRetryStrategy struct {
backoffCalculator BackoffCalculator
}
// NewBestEffortRetryStrategy returns a new BestEffortRetryStrategy which will use the supplied calculator function
// to calculate retry durations. If calculator is nil then ControlledBackoff will be used.
func NewBestEffortRetryStrategy(calculator BackoffCalculator) *BestEffortRetryStrategy {
if calculator == nil {
calculator = ControlledBackoff
}
return &BestEffortRetryStrategy{backoffCalculator: calculator}
}
// RetryAfter calculates and returns a RetryAction describing how long to wait before retrying an operation.
func (rs *BestEffortRetryStrategy) RetryAfter(req RetryRequest, reason RetryReason) RetryAction {
if req.Idempotent() || reason.AllowsNonIdempotentRetry() {
return &WithDurationRetryAction{WithDuration: rs.backoffCalculator(req.RetryAttempts())}
}
return &NoRetryRetryAction{}
}
// ExponentialBackoff calculates a backoff time duration from the retry attempts on a given request.
func ExponentialBackoff(min, max time.Duration, backoffFactor float64) BackoffCalculator {
var minBackoff float64 = 1000000 // 1 Millisecond
var maxBackoff float64 = 500000000 // 500 Milliseconds
var factor float64 = 2
if min > 0 {
minBackoff = float64(min)
}
if max > 0 {
maxBackoff = float64(max)
}
if backoffFactor > 0 {
factor = backoffFactor
}
return func(retryAttempts uint32) time.Duration {
backoff := minBackoff * (math.Pow(factor, float64(retryAttempts)))
if backoff > maxBackoff {
backoff = maxBackoff
}
if backoff < minBackoff {
backoff = minBackoff
}
return time.Duration(backoff)
}
}
// ControlledBackoff calculates a backoff time duration from the retry attempts on a given request.
func ControlledBackoff(retryAttempts uint32) time.Duration {
switch retryAttempts {
case 0:
return 1 * time.Millisecond
case 1:
return 10 * time.Millisecond
case 2:
return 50 * time.Millisecond
case 3:
return 100 * time.Millisecond
case 4:
return 500 * time.Millisecond
default:
return 1000 * time.Millisecond
}
}
var idempotentOps = map[memd.CmdCode]bool{
memd.CmdGet: true,
memd.CmdGetReplica: true,
memd.CmdGetMeta: true,
memd.CmdSubDocGet: true,
memd.CmdSubDocExists: true,
memd.CmdSubDocGetCount: true,
memd.CmdNoop: true,
memd.CmdStat: true,
memd.CmdGetRandom: true,
memd.CmdCollectionsGetID: true,
memd.CmdCollectionsGetManifest: true,
memd.CmdGetClusterConfig: true,
memd.CmdObserve: true,
memd.CmdObserveSeqNo: true,
}