Prometheusサーバで収集されたデータの取り出し方として、PrometheusはHTTP APIを提供しています。
Prometheus HTTP API
Exporterをカスタマイズして作成する方法はたくさん記事がありましたが、PrometheusのデータをHTTP APIから取得する方法についての情報があまり見当たらなかったので備忘のため記録しておきます。
Prometheus HTTP APIをcurlで呼び出してみる
prometheusに登録されているtarget一覧の取得
$ curl -G "http://<prometheus ホスト:ポート>/api/v1/targets" | jq .
{
"status": "success",
"data": {
"activeTargets": [
{
"discoveredLabels": {
"__address__": "192.168.212.101:8001",
"__metrics_path__": "/metrics",
"__scheme__": "http",
"job": "kong"
},
"labels": {
"instance": "192.168.212.101:8001",
"job": "kong"
},
"scrapeUrl": "http://192.168.212.101:8001/metrics",
"lastError": "",
"lastScrape": "2019-10-29T11:07:40.26459926Z",
"health": "up"
},
{
"discoveredLabels": {
"__address__": "localhost:9090",
"__metrics_path__": "/metrics",
"__scheme__": "http",
"job": "prometheus"
},
"labels": {
"instance": "localhost:9090",
"job": "prometheus"
},
"scrapeUrl": "http://localhost:9090/metrics",
"lastError": "",
"lastScrape": "2019-10-29T11:07:33.962713961Z",
"health": "up"
}
],
"droppedTargets": []
}
}
ターゲットのメトリクス一覧の情報取得
job名としてkongと設定したtargetのメタデータを取得する例です。(kong api gatewayのメトリクスリストが取れているのがわかります。)
$ curl -G 'http://<prometheus ホスト:ポート>/api/v1/targets/metadata' --data-urlencode 'match_target={job="kong"}' | jq .
{
"status": "success",
"data": [
{
"target": {
"instance": "192.168.212.101:8001",
"job": "kong"
},
"metric": "kong_nginx_http_current_connections",
"type": "gauge",
"help": "Number of HTTP connections",
"unit": ""
},
{
"target": {
"instance": "192.168.212.101:8001",
"job": "kong"
},
"metric": "kong_nginx_metric_errors_total",
"type": "counter",
"help": "Number of nginx-lua-prometheus errors",
"unit": ""
},
{
"target": {
"instance": "192.168.212.101:8001",
"job": "kong"
},
"metric": "kong_bandwidth",
"type": "counter",
"help": "Total bandwidth in bytes consumed per service in Kong",
"unit": ""
},
{
"target": {
"instance": "192.168.212.101:8001",
"job": "kong"
},
"metric": "kong_datastore_reachable",
"type": "gauge",
"help": "Datastore reachable from Kong, 0 is unreachable",
"unit": ""
},
{
"target": {
"instance": "192.168.212.101:8001",
"job": "kong"
},
"metric": "kong_http_status",
"type": "counter",
"help": "HTTP status codes per service in Kong",
"unit": ""
},
{
"target": {
"instance": "192.168.212.101:8001",
"job": "kong"
},
"metric": "kong_latency",
"type": "histogram",
"help": "Latency added by Kong, total request time and upstream latency for each service in Kong",
"unit": ""
}
]
}
メトリクスのデータを取得
kong_nginx_http_current_connectionsメトリクスの値に対して、Tue 29 Oct 2019 08:36:42 PM JSTからTue 29 Oct 2019 08:46:42 PM JSTの間の10分間のデータを1分間隔で集計した時系列データを取得する例。
$ curl -G http://<prometheus ホスト:ポート>/api/v1/query_range --data-urlencode 'query=kong_nginx_http_current_connections' -d 'start=1572349002' -d 'end=1572349602'
-d 'step=60s' | jq .
{
"status": "success",
"data": {
"resultType": "matrix",
"result": [
{
"metric": {
"__name__": "kong_nginx_http_current_connections",
"instance": "192.168.212.101:8001",
"job": "kong",
"state": "accepted"
},
"values": [
[
1572349002,
"6"
],
[
1572349062,
"6"
],
[
1572349122,
"6"
],
[
1572349182,
"6"
],
[
1572349242,
"6"
],
[
1572349302,
"6"
],
[
1572349362,
"6"
],
[
1572349422,
"6"
],
[
1572349482,
"6"
],
[
1572349542,
"6"
],
[
1572349602,
"6"
]
]
},
{
"metric": {
"__name__": "kong_nginx_http_current_connections",
"instance": "192.168.212.101:8001",
"job": "kong",
"state": "active"
},
"values": [
[
1572349002,
"3"
],
[
1572349062,
"3"
],
[
1572349122,
"3"
],
[
1572349182,
"3"
],
[
1572349242,
"3"
],
[
1572349302,
"3"
],
[
1572349362,
"3"
],
[
1572349422,
"3"
],
[
1572349482,
"3"
],
[
1572349542,
"3"
],
[
1572349602,
"3"
]
]
},
{
"metric": {
"__name__": "kong_nginx_http_current_connections",
"instance": "192.168.212.101:8001",
"job": "kong",
"state": "handled"
},
"values": [
[
1572349002,
"6"
],
[
1572349062,
"6"
],
[
1572349122,
"6"
],
[
1572349182,
"6"
],
[
1572349242,
"6"
],
[
1572349302,
"6"
],
[
1572349362,
"6"
],
[
1572349422,
"6"
],
[
1572349482,
"6"
],
[
1572349542,
"6"
],
[
1572349602,
"6"
]
]
},
・・・略
}
]
}
}
Prometheusのgolang clientライブラリを使って取得してみる
ライブラリを使うにはまず以下のようにgo getで取得しておきます。
$ go get github.com/prometheus/client_golang/api
$ go get github.com/prometheus/client_golang/api/prometheus/v1
上記のcurlで行っている処理と同様の処理をgolangで取得してみます。
prometheus_api_sample.go
package main
import (
"github.com/prometheus/client_golang/api"
v1 "github.com/prometheus/client_golang/api/prometheus/v1"
"fmt"
"os"
"time"
"context"
)
func main() {
client, err := api.NewClient(api.Config{
Address: "http://192.168.212.101:9099",
})
if err != nil {
fmt.Printf("Error creating client: %v\n", err)
os.Exit(1)
}
api := v1.NewAPI(client)
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
defer cancel()
// /api/v1/targets
result, err := api.Targets(ctx)
if err != nil {
fmt.Printf("Error get Targets: %v\n", err)
os.Exit(1)
}
fmt.Printf("Result:\n%v\n", result)
// /api/v1/target/metadata
result_metric_metadata, err := api.TargetsMetadata(ctx, "{job=\"kong\"}", "", "")
if err != nil {
fmt.Printf("Error get target metadata: %v\n", err)
os.Exit(1)
}
fmt.Printf("Result TargetsMetadata: \n%v\n", result_metric_metadata)
// /api/v1/query_range
query := "kong_nginx_http_current_connections"
now := time.Now()
range_param := v1.Range{Start: now.Add(-time.Hour), End: now, Step: 60 * time.Second}
result_query_range, warning, err := api.QueryRange(ctx, query, range_param)
if err != nil {
fmt.Printf("Error get query range: %v\n", err)
os.Exit(1)
}
if len(warning) > 0 {
fmt.Printf("Warning QueryRange: %v\n", warning)
}
fmt.Printf("Result QueryRange: \n%v\n", result_query_range)
}
実行してみると以下のような結果になります。
$ go run prometheus_api_sample.go
Result:
{[{map[__address__:192.168.212.101:8001 __metrics_path__:/metrics __scheme__:http job:kong] {instance="192.168.212.101:8001", job="kong"} http://192.168.212.101:8001/metrics 2019-10-29
15:06:10.892269867 +0000 UTC up} {map[__address__:localhost:9090 __metrics_path__:/metrics __scheme__:http job:prometheus] {instance="localhost:9090", job="prometheus"} http://localhost:9090/metrics 2019-10-29 15:06:19.589531881 +0000 UTC up}] []}
Result TargetsMetadata:
[{map[instance:192.168.212.101:8001 job:kong] kong_nginx_http_current_connections gauge Number of HTTP connections } {map[instance:192.168.212.101:8001 job:kong] kong_nginx_metric_errors_total counter Number of nginx-lua-prometheus errors } {map[instance:192.168.212.101:8001 job:kong] kong_bandwidth counter Total bandwidth in bytes consumed per service in Kong } {map[instance:192.168.212.101:8001 job:kong] kong_datastore_reachable gauge Datastore reachable from Kong, 0 is unreachable } {map[instance:192.168.212.101:8001 job:kong] kong_http_status counter HTTP status codes per service in Kong } {map[instance:192.168.212.101:8001 job:kong] kong_latency histogram Latency added by Kong, total request time and upstream latency for each service in Kong }]
Result QueryRange:
kong_nginx_http_current_connections{instance="192.168.212.101:8001", job="kong", state="accepted"} =>
14 @[1572358764.66]
14 @[1572358824.66]
14 @[1572358884.66]
・・・略
17 @[1572361464.66]
17 @[1572361524.66]
17 @[1572361584.66]
kong_nginx_http_current_connections{instance="192.168.212.101:8001", job="kong", state="active"} =>
3 @[1572358764.66]
3 @[1572358824.66]
・・・略
3 @[1572359724.66]
3 @[1572359784.66]
・・・略