-
Notifications
You must be signed in to change notification settings - Fork 239
/
gcp_exporter.go
231 lines (203 loc) · 7.48 KB
/
gcp_exporter.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
package gcp_exporter
import (
"context"
"crypto/md5"
"encoding/hex"
"errors"
"fmt"
"net/http"
"strings"
"time"
"github.com/PuerkitoBio/rehttp"
"github.com/go-kit/log"
"github.com/grafana/dskit/multierror"
"github.com/prometheus-community/stackdriver_exporter/collectors"
"github.com/prometheus-community/stackdriver_exporter/delta"
"github.com/prometheus-community/stackdriver_exporter/utils"
"github.com/prometheus/client_golang/prometheus"
"golang.org/x/oauth2/google"
"google.golang.org/api/monitoring/v3"
"google.golang.org/api/option"
"gopkg.in/yaml.v2"
"github.com/grafana/alloy/internal/runtime/logging/level"
"github.com/grafana/alloy/internal/static/integrations"
integrations_v2 "github.com/grafana/alloy/internal/static/integrations/v2"
"github.com/grafana/alloy/internal/static/integrations/v2/metricsutils"
)
func init() {
integrations.RegisterIntegration(&Config{})
integrations_v2.RegisterLegacy(&Config{}, integrations_v2.TypeMultiplex, metricsutils.NewNamedShim("gcp"))
}
type Config struct {
ProjectIDs []string `yaml:"project_ids"`
MetricPrefixes []string `yaml:"metrics_prefixes"`
ExtraFilters []string `yaml:"extra_filters"`
RequestInterval time.Duration `yaml:"request_interval"`
RequestOffset time.Duration `yaml:"request_offset"`
IngestDelay bool `yaml:"ingest_delay"`
DropDelegatedProjects bool `yaml:"drop_delegated_projects"`
ClientTimeout time.Duration `yaml:"gcp_client_timeout"`
}
var DefaultConfig = Config{
ClientTimeout: 15 * time.Second,
RequestInterval: 5 * time.Minute,
RequestOffset: 0,
IngestDelay: false,
DropDelegatedProjects: false,
}
// UnmarshalYAML implements yaml.Unmarshaler for Config
func (c *Config) UnmarshalYAML(unmarshal func(interface{}) error) error {
*c = DefaultConfig
type plain Config
return unmarshal((*plain)(c))
}
func (c *Config) Name() string {
return "gcp_exporter"
}
func (c *Config) InstanceKey(_ string) (string, error) {
// We use a hash of the config so our key is unique when leveraged with v2
// The config itself doesn't have anything which can uniquely identify it.
bytes, err := yaml.Marshal(c)
if err != nil {
return "", err
}
hash := md5.Sum(bytes)
return hex.EncodeToString(hash[:]), nil
}
func (c *Config) NewIntegration(l log.Logger) (integrations.Integration, error) {
if err := c.Validate(); err != nil {
return nil, err
}
svc, err := createMonitoringService(context.Background(), c.ClientTimeout)
if err != nil {
return nil, err
}
var gcpCollectors []prometheus.Collector
var counterStores []*SelfPruningDeltaStore[collectors.ConstMetric]
var histogramStores []*SelfPruningDeltaStore[collectors.HistogramMetric]
for _, projectID := range c.ProjectIDs {
counterStore := NewSelfPruningDeltaStore[collectors.ConstMetric](l, delta.NewInMemoryCounterStore(l, 30*time.Minute))
histogramStore := NewSelfPruningDeltaStore[collectors.HistogramMetric](l, delta.NewInMemoryHistogramStore(l, 30*time.Minute))
monitoringCollector, err := collectors.NewMonitoringCollector(
projectID,
svc,
collectors.MonitoringCollectorOptions{
MetricTypePrefixes: c.MetricPrefixes,
ExtraFilters: parseMetricExtraFilters(c.ExtraFilters),
RequestInterval: c.RequestInterval,
RequestOffset: c.RequestOffset,
IngestDelay: c.IngestDelay,
DropDelegatedProjects: c.DropDelegatedProjects,
// If FillMissingLabels ensures all metrics with the same name have the same label set. It's not often
// that GCP metrics have different labels but if it happens the prom registry will panic.
FillMissingLabels: true,
// If AggregateDeltas is disabled the data produced is not useful at all. See https://github.com/prometheus-community/stackdriver_exporter#what-to-know-about-aggregating-delta-metrics
// for more info
AggregateDeltas: true,
},
l,
counterStore,
histogramStore,
)
if err != nil {
return nil, fmt.Errorf("failed to create monitoring collector: %w", err)
}
counterStores = append(counterStores, counterStore)
histogramStores = append(histogramStores, histogramStore)
gcpCollectors = append(gcpCollectors, monitoringCollector)
}
run := func(ctx context.Context) error {
ticker := time.NewTicker(5 * time.Minute)
defer ticker.Stop()
for {
select {
case <-ticker.C:
level.Debug(l).Log("msg", "Starting delta store pruning", "number_of_stores", len(counterStores)+len(histogramStores))
for _, store := range counterStores {
store.Prune(ctx)
}
for _, store := range histogramStores {
store.Prune(ctx)
}
level.Debug(l).Log("msg", "Finished delta store pruning", "number_of_stores", len(counterStores)+len(histogramStores))
case <-ctx.Done():
return ctx.Err()
}
}
}
return integrations.NewCollectorIntegration(
c.Name(), integrations.WithCollectors(gcpCollectors...), integrations.WithRunner(run),
), nil
}
func (c *Config) Validate() error {
configErrors := multierror.MultiError{}
if c.ProjectIDs == nil || len(c.ProjectIDs) == 0 {
configErrors.Add(errors.New("no project_ids defined"))
}
if c.MetricPrefixes == nil || len(c.MetricPrefixes) == 0 {
configErrors.Add(errors.New("at least 1 metrics_prefixes is required"))
}
if len(c.ExtraFilters) > 0 {
filterPrefixToFilter := map[string][]string{}
for _, filter := range c.ExtraFilters {
splitFilter := strings.Split(filter, ":")
if len(splitFilter) <= 1 {
configErrors.Add(fmt.Errorf("%s is an invalid filter a filter must be of the form <metric_type>:<filter_expression>", filter))
continue
}
filterPrefix := splitFilter[0]
if _, exists := filterPrefixToFilter[filterPrefix]; !exists {
filterPrefixToFilter[filterPrefix] = []string{}
}
filterPrefixToFilter[filterPrefix] = append(filterPrefixToFilter[filterPrefix], filter)
}
for filterPrefix, filters := range filterPrefixToFilter {
validFilterPrefix := false
for _, metricPrefix := range c.MetricPrefixes {
if strings.HasPrefix(metricPrefix, filterPrefix) {
validFilterPrefix = true
break
}
}
if !validFilterPrefix {
configErrors.Add(fmt.Errorf("no metric_prefixes started with %s which means the extra_filters %s will not have any effect", filterPrefix, strings.Join(filters, ",")))
}
}
}
return configErrors.Err()
}
func createMonitoringService(ctx context.Context, httpTimeout time.Duration) (*monitoring.Service, error) {
googleClient, err := google.DefaultClient(ctx, monitoring.MonitoringReadScope)
if err != nil {
return nil, fmt.Errorf("error creating Google client: %v", err)
}
googleClient.Timeout = httpTimeout
googleClient.Transport = rehttp.NewTransport(
googleClient.Transport,
rehttp.RetryAll(
rehttp.RetryMaxRetries(4),
rehttp.RetryStatuses(http.StatusServiceUnavailable)),
rehttp.ExpJitterDelay(time.Second, 5*time.Second),
)
monitoringService, err := monitoring.NewService(ctx,
option.WithHTTPClient(googleClient),
)
if err != nil {
return nil, fmt.Errorf("error creating Google Stackdriver Monitoring service: %v", err)
}
return monitoringService, nil
}
func parseMetricExtraFilters(filters []string) []collectors.MetricFilter {
var extraFilters []collectors.MetricFilter
for _, ef := range filters {
efPrefix, efModifier := utils.GetExtraFilterModifiers(ef, ":")
if efPrefix != "" {
extraFilter := collectors.MetricFilter{
Prefix: efPrefix,
Modifier: efModifier,
}
extraFilters = append(extraFilters, extraFilter)
}
}
return extraFilters
}