-
Notifications
You must be signed in to change notification settings - Fork 754
/
telemetry.go
292 lines (256 loc) · 10 KB
/
telemetry.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
// The MIT License
//
// Copyright (c) 2020 Temporal Technologies Inc. All rights reserved.
//
// Copyright (c) 2020 Uber Technologies, Inc.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.
package interceptor
import (
"context"
"time"
"go.temporal.io/api/enums/v1"
"go.temporal.io/api/serviceerror"
"go.temporal.io/api/workflowservice/v1"
"google.golang.org/grpc"
"go.temporal.io/server/common"
"go.temporal.io/server/common/log"
"go.temporal.io/server/common/log/tag"
"go.temporal.io/server/common/metrics"
"go.temporal.io/server/common/namespace"
serviceerrors "go.temporal.io/server/common/serviceerror"
)
type (
metricsContextKey struct{}
TelemetryInterceptor struct {
namespaceRegistry namespace.Registry
metricsClient metrics.Client
scopes map[string]int
logger log.Logger
}
)
var (
metricsCtxKey = metricsContextKey{}
_ grpc.UnaryServerInterceptor = (*TelemetryInterceptor)(nil).Intercept
)
// static variables used to emit action metrics.
var (
respondWorkflowTaskCompleted = "RespondWorkflowTaskCompleted"
pollActivityTaskQueue = "PollActivityTaskQueue"
grpcActions = map[string]int{
"QueryWorkflow": metrics.FrontendQueryWorkflowScope,
"RecordActivityTaskHeartbeat": metrics.FrontendRecordActivityTaskHeartbeatScope,
"RecordActivityTaskHeartbeatById": metrics.FrontendRecordActivityTaskHeartbeatByIdScope,
"ResetWorkflowExecution": metrics.FrontendResetWorkflowExecutionScope,
"StartWorkflowExecution": metrics.FrontendStartWorkflowExecutionScope,
"SignalWorkflowExecution": metrics.FrontendSignalWorkflowExecutionScope,
"SignalWithStartWorkflowExecution": metrics.FrontendSignalWithStartWorkflowExecutionScope,
respondWorkflowTaskCompleted: metrics.FrontendRespondWorkflowTaskCompletedScope,
pollActivityTaskQueue: metrics.FrontendPollActivityTaskQueueScope,
}
commandActions = map[enums.CommandType]struct{}{
enums.COMMAND_TYPE_RECORD_MARKER: {},
enums.COMMAND_TYPE_START_TIMER: {},
enums.COMMAND_TYPE_SCHEDULE_ACTIVITY_TASK: {},
enums.COMMAND_TYPE_START_CHILD_WORKFLOW_EXECUTION: {},
enums.COMMAND_TYPE_SIGNAL_EXTERNAL_WORKFLOW_EXECUTION: {},
enums.COMMAND_TYPE_UPSERT_WORKFLOW_SEARCH_ATTRIBUTES: {},
enums.COMMAND_TYPE_MODIFY_WORKFLOW_PROPERTIES: {},
enums.COMMAND_TYPE_CONTINUE_AS_NEW_WORKFLOW_EXECUTION: {},
}
)
func NewTelemetryInterceptor(
namespaceRegistry namespace.Registry,
metricsClient metrics.Client,
scopes map[string]int,
logger log.Logger,
) *TelemetryInterceptor {
return &TelemetryInterceptor{
namespaceRegistry: namespaceRegistry,
metricsClient: metricsClient,
scopes: scopes,
logger: logger,
}
}
// Use this method to override scope used for reporting a metric.
// Ideally this method should never be used.
func (ti *TelemetryInterceptor) overrideScope(scope int, req interface{}) int {
// GetWorkflowExecutionHistory method handles both long poll and regular calls.
// Current plan is to eventually split GetWorkflowExecutionHistory into two APIs,
// remove this "if" case when that is done.
if scope == metrics.FrontendGetWorkflowExecutionHistoryScope {
request := req.(*workflowservice.GetWorkflowExecutionHistoryRequest)
if request.GetWaitNewEvent() {
return metrics.FrontendPollWorkflowExecutionHistoryScope
}
}
return scope
}
func (ti *TelemetryInterceptor) Intercept(
ctx context.Context,
req interface{},
info *grpc.UnaryServerInfo,
handler grpc.UnaryHandler,
) (interface{}, error) {
_, methodName := splitMethodName(info.FullMethod)
metricsScope, logTags := ti.metricsScopeLogTags(req, info.FullMethod, methodName)
ctx = context.WithValue(ctx, metricsCtxKey, metricsScope)
metricsScope.IncCounter(metrics.ServiceRequests)
timer := metricsScope.StartTimer(metrics.ServiceLatency)
defer timer.Stop()
timerNoUserLatency := metricsScope.StartTimer(metrics.ServiceLatencyNoUserLatency)
defer timerNoUserLatency.Stop()
resp, err := handler(ctx, req)
if val, ok := metrics.ContextCounterGet(ctx, metrics.HistoryWorkflowExecutionCacheLatency); ok {
userLatencyDuration := time.Duration(val)
timerNoUserLatency.Subtract(userLatencyDuration)
metricsScope.RecordTimer(metrics.ServiceLatencyUserLatency, userLatencyDuration)
}
if err != nil {
ti.handleError(metricsScope, logTags, err)
return nil, err
}
// emit action metrics only after successful calls
ti.emitActionMetric(methodName, info.FullMethod, req, metricsScope, resp)
return resp, nil
}
func (ti *TelemetryInterceptor) emitActionMetric(
methodName string,
fullName string,
req interface{},
scope metrics.Scope,
result interface{},
) {
if actionScope, ok := grpcActions[methodName]; !ok || actionScope != ti.scopes[fullName] {
// grpcActions checks that methodName is the one that we care about.
// ti.scopes verifies that the scope is the one we intended to emit action metrics.
// This is necessary because TelemetryInterceptor is used for all services. Different service could have same
// method name. But we only want to emit action metrics from frontend.
return
}
switch methodName {
case respondWorkflowTaskCompleted:
// handle commands
completedRequest, ok := req.(*workflowservice.RespondWorkflowTaskCompletedRequest)
if !ok {
return
}
for _, command := range completedRequest.Commands {
if _, ok := commandActions[command.CommandType]; ok {
switch command.CommandType {
case enums.COMMAND_TYPE_RECORD_MARKER:
// handle RecordMarker command, they are used for localActivity, sideEffect, versioning etc.
markerName := command.GetRecordMarkerCommandAttributes().GetMarkerName()
scope.Tagged(metrics.ActionType("command_RecordMarker_" + markerName)).IncCounter(metrics.ActionCounter)
default:
// handle all other command action
scope.Tagged(metrics.ActionType("command_" + command.CommandType.String())).IncCounter(metrics.ActionCounter)
}
}
}
case pollActivityTaskQueue:
// handle activity retries
activityPollResponse, ok := result.(*workflowservice.PollActivityTaskQueueResponse)
if !ok {
return
}
if activityPollResponse == nil || len(activityPollResponse.TaskToken) == 0 {
// empty response
return
}
if activityPollResponse.Attempt > 1 {
scope.Tagged(metrics.ActionType("activity_retry")).IncCounter(metrics.ActionCounter)
}
default:
// grpc action
scope.Tagged(metrics.ActionType("grpc_" + methodName)).IncCounter(metrics.ActionCounter)
}
}
func (ti *TelemetryInterceptor) metricsScopeLogTags(
req interface{},
fullMethod string,
methodName string,
) (metrics.Scope, []tag.Tag) {
// if the method name is not defined, will default to
// unknown scope, which carries value 0
scopeDef := ti.scopes[fullMethod]
scopeDef = ti.overrideScope(scopeDef, req)
nsName := GetNamespace(ti.namespaceRegistry, req)
if nsName == "" {
return ti.metricsClient.Scope(scopeDef).Tagged(metrics.NamespaceUnknownTag()), []tag.Tag{tag.Operation(methodName)}
}
return ti.metricsClient.Scope(scopeDef).Tagged(metrics.NamespaceTag(nsName.String())), []tag.Tag{
tag.Operation(methodName),
tag.WorkflowNamespace(nsName.String()),
}
}
func (ti *TelemetryInterceptor) handleError(
scope metrics.Scope,
logTags []tag.Tag,
err error,
) {
scope.Tagged(metrics.ServiceErrorTypeTag(err)).IncCounter(metrics.ServiceErrorWithType)
if common.IsContextDeadlineExceededErr(err) || common.IsContextCanceledErr(err) {
return
}
switch err := err.(type) {
// we emit service_error_with_type metrics, no need to emit specific metric for these known error types.
case *serviceerror.AlreadyExists,
*serviceerror.CancellationAlreadyRequested,
*serviceerror.NamespaceInvalidState,
*serviceerror.NamespaceNotActive,
*serviceerror.NamespaceNotFound,
*serviceerror.NamespaceAlreadyExists,
*serviceerror.InvalidArgument,
*serviceerror.WorkflowExecutionAlreadyStarted,
*serviceerror.WorkflowNotReady,
*serviceerror.NotFound,
*serviceerror.QueryFailed,
*serviceerror.ClientVersionNotSupported,
*serviceerror.ServerVersionNotSupported,
*serviceerror.PermissionDenied,
*serviceerrors.StickyWorkerUnavailable,
*serviceerrors.ShardOwnershipLost,
*serviceerrors.TaskAlreadyStarted,
*serviceerrors.RetryReplication:
// no-op
// specific metric for resource exhausted error with throttle reason
case *serviceerror.ResourceExhausted:
scope.Tagged(metrics.ResourceExhaustedCauseTag(err.Cause)).IncCounter(metrics.ServiceErrResourceExhaustedCounter)
// Any other errors are treated as ServiceFailures against SLA.
// Including below known errors and any other unknown errors.
// *serviceerror.DataLoss,
// *serviceerror.Internal
// *serviceerror.Unavailable:
default:
scope.IncCounter(metrics.ServiceFailures)
ti.logger.Error("service failures", append(logTags, tag.Error(err))...)
}
}
func MetricsScope(
ctx context.Context,
logger log.Logger,
) metrics.Scope {
scope, ok := ctx.Value(metricsCtxKey).(metrics.Scope)
if !ok {
logger.Error("unable to get metrics scope")
return metrics.NoopScope
}
return scope
}