Add pgsql scenario

api/buf.lock

@@ -9,8 +9,8 @@ deps:
   - remote: buf.build
     owner: envoyproxy
     repository: envoy
-    commit: 5b59168cb47246a3b40d27598b6d5909
-    digest: shake256:e219b11f3e40c9e2956286c5de5d2cb3ea6288228b5f6126be163dbe62b050e7accef81d13e5cf35b447e6903a3f21b82019e676ad8b858ca60c48ca8126e391
+    commit: 39f01805b8754956a78bc77bda4c3761
+    digest: shake256:a7e7c7675618056879597d6205e394e85b63b71fd1766818042e6fd9b5c4fd623bede2784a9c635dced61fb40a22610812f164ed40c7a0cecbe50c0ecc76952d
   - remote: buf.build
     owner: envoyproxy
     repository: protoc-gen-validate

blueprints/concurrency-limiting/base/config.libsonnet

@@ -13,7 +13,6 @@ commonConfig {
       max_concurrency: '__REQUIRED_FIELD__',
       selectors: commonConfig.selectors_defaults,
       parameters: {
-        limit_by_label_key: 'limit_by_label_key',
         max_inflight_duration: '__REQUIRED_FIELD__',
       },
       request_parameters: {},

blueprints/concurrency-limiting/base/gen/definitions.json

@@ -67,7 +67,6 @@
             "parameters": {
               "description": "Parameters.",
               "default": {
-                "limit_by_label_key": "limit_by_label_key",
                 "max_inflight_duration": "__REQUIRED_FIELD__"
               },
               "type": "object",

blueprints/concurrency-limiting/base/gen/values.yaml

@@ -29,7 +29,6 @@ policy:
     # Type: aperture.spec.v1.ConcurrencyLimiterParameters
     # Required: True
     parameters:
-      limit_by_label_key: "limit_by_label_key"
       max_inflight_duration: __REQUIRED_FIELD__
     # Request Parameters.
     # Type: aperture.spec.v1.ConcurrencyLimiterRequestParameters

blueprints/concurrency-scheduling/base/config.libsonnet

@@ -13,14 +13,9 @@ commonConfig {
       max_concurrency: '__REQUIRED_FIELD__',
       selectors: commonConfig.selectors_defaults,
       concurrency_limiter: {
-        limit_by_label_key: 'limit_by_label_key',
         max_inflight_duration: '__REQUIRED_FIELD__',
       },
-      scheduler: {
-        tokens_label_key: 'tokens',
-        priority_label_key: 'priority',
-        workload_label_key: 'workload',
-      },
+      scheduler: {},
       alerter: {
         alert_name: 'Too many inflight requests',
       },

blueprints/concurrency-scheduling/base/gen/definitions.json

@@ -61,7 +61,6 @@
             "concurrency_limiter": {
               "description": "Concurrency Limiter Parameters.",
               "default": {
-                "limit_by_label_key": "limit_by_label_key",
                 "max_inflight_duration": "__REQUIRED_FIELD__"
               },
               "type": "object",
@@ -75,11 +74,7 @@
             },
             "scheduler": {
               "description": "Scheduler configuration.",
-              "default": {
-                "priority_label_key": "priority",
-                "tokens_label_key": "tokens",
-                "workload_label_key": "workload"
-              },
+              "default": {},
               "type": "object",
               "$ref": "../../../gen/jsonschema/_definitions.json#/definitions/Scheduler"
             },

blueprints/concurrency-scheduling/base/gen/values.yaml

@@ -25,18 +25,14 @@ policy:
     # Type: aperture.spec.v1.ConcurrencyLimiterParameters
     # Required: True
     concurrency_limiter:
-      limit_by_label_key: "limit_by_label_key"
       max_inflight_duration: __REQUIRED_FIELD__
     # Max concurrency.
     # Type: float64
     # Required: True
     max_concurrency: __REQUIRED_FIELD__
     # Scheduler configuration.
     # Type: aperture.spec.v1.Scheduler
-    scheduler:
-      priority_label_key: "priority"
-      tokens_label_key: "tokens"
-      workload_label_key: "workload"
+    scheduler: {}
     # Flow selectors to match requests against.
     # Type: []aperture.spec.v1.Selector
     # Required: True

blueprints/quota-scheduling/base/config.libsonnet

@@ -15,14 +15,9 @@ commonConfig {
       fill_amount: '__REQUIRED_FIELD__',
       selectors: commonConfig.selectors_defaults,
       rate_limiter: {
-        limit_by_label_key: 'limit_key',
         interval: '__REQUIRED_FIELD__',
       },
-      scheduler: {
-        tokens_label_key: 'tokens',
-        priority_label_key: 'priority',
-        workload_label_key: 'workload',
-      },
+      scheduler: {},
       alerter: {
         alert_name: 'More than 90% of requests are being rate limited',
       },

blueprints/quota-scheduling/base/gen/definitions.json

@@ -78,19 +78,14 @@
             "rate_limiter": {
               "description": "Rate Limiter Parameters.",
               "default": {
-                "interval": "__REQUIRED_FIELD__",
-                "limit_by_label_key": "limit_key"
+                "interval": "__REQUIRED_FIELD__"
               },
               "type": "object",
               "$ref": "../../../gen/jsonschema/_definitions.json#/definitions/RateLimiterParameters"
             },
             "scheduler": {
               "description": "Scheduler configuration.",
-              "default": {
-                "priority_label_key": "priority",
-                "tokens_label_key": "tokens",
-                "workload_label_key": "workload"
-              },
+              "default": {},
               "type": "object",
               "$ref": "../../../gen/jsonschema/_definitions.json#/definitions/Scheduler"
             },

blueprints/quota-scheduling/base/gen/values.yaml

@@ -34,13 +34,9 @@ policy:
     # Required: True
     rate_limiter:
       interval: __REQUIRED_FIELD__
-      limit_by_label_key: "limit_key"
     # Scheduler configuration.
     # Type: aperture.spec.v1.Scheduler
-    scheduler:
-      priority_label_key: "priority"
-      tokens_label_key: "tokens"
-      workload_label_key: "workload"
+    scheduler: {}
     # Flow selectors to match requests against.
     # Type: []aperture.spec.v1.Selector
     # Required: True

blueprints/rate-limiting/base/config.libsonnet

@@ -15,7 +15,6 @@ commonConfig {
       fill_amount: '__REQUIRED_FIELD__',
       selectors: commonConfig.selectors_defaults,
       parameters: {
-        limit_by_label_key: 'limit_key',
         interval: '__REQUIRED_FIELD__',
       },
       request_parameters: {},

blueprints/rate-limiting/base/gen/definitions.json

@@ -78,8 +78,7 @@
             "parameters": {
               "description": "Parameters.",
               "default": {
-                "interval": "__REQUIRED_FIELD__",
-                "limit_by_label_key": "limit_key"
+                "interval": "__REQUIRED_FIELD__"
               },
               "type": "object",
               "$ref": "../../../gen/jsonschema/_definitions.json#/definitions/RateLimiterParameters"

blueprints/rate-limiting/base/gen/values.yaml

@@ -34,7 +34,6 @@ policy:
     # Required: True
     parameters:
       interval: __REQUIRED_FIELD__
-      limit_by_label_key: "limit_key"
     # Request Parameters.
     # Type: aperture.spec.v1.RateLimiterRequestParameters
     request_parameters: {}

docs/content/aperture-for-infra/guides/api-quota-management/assets/inter-service-rate-limiting/policy.yaml

@@ -26,9 +26,6 @@ spec:
               num_sync: 4
             limit_by_label_key: api_key
           scheduler:
-            priority_label_key: priority
-            tokens_label_key: tokens
-            workload_label_key: workload
             workloads:
             - label_matcher:
                 match_labels:

docs/content/guides/assets/managing-quotas/policy.yaml

@@ -24,7 +24,6 @@ spec:
             limit_by_label_key: user_id
           scheduler:
             priority_label_key: priority
-            tokens_label_key: tokens
             workload_label_key: workload
           selectors:
           - control_point: quota-scheduling-feature

docs/content/guides/assets/openai/policy.yaml

@@ -24,8 +24,6 @@ spec:
             limit_by_label_key: api_key
           scheduler:
             priority_label_key: priority
-            tokens_label_key: tokens
-            workload_label_key: workload
           selectors:
           - control_point: openai
             label_matcher:

docs/content/reference/aperture-cli/policies/assets/raw_values.yaml

@@ -34,7 +34,6 @@ policy:
     # Required: True
     parameters:
       interval: __REQUIRED_FIELD__
-      limit_by_label_key: "limit_key"
     # Request Parameters.
     # Type: aperture.spec.v1.RateLimiterRequestParameters
     request_parameters: {}

docs/content/reference/blueprints/concurrency-limiting/base.md

@@ -9,7 +9,7 @@
 
 This blueprint provides a concurrency limiting policy and a dashboard. This
 policy uses the
 [`ConcurrencyLimiter`](/reference/configuration/spec.md#concurrency-limiter)
 component.
 
 <!-- Configuration Marker -->
@@ -121,7 +121,7 @@
     description='Parameters.'
     type='Object (aperture.spec.v1.ConcurrencyLimiterParameters)'
     reference='../../configuration/spec#concurrency-limiter-parameters'
-    value='{"limit_by_label_key": "limit_by_label_key", "max_inflight_duration": "__REQUIRED_FIELD__"}'
+    value='{"max_inflight_duration": "__REQUIRED_FIELD__"}'
 />
 
 <!-- vale on -->

docs/content/reference/blueprints/concurrency-scheduling/base.md

@@ -9,7 +9,7 @@
 
 This blueprint provides a concurrency scheduling policy and a dashboard. This
 policy uses the
 [`ConcurrencyScheduler`](/reference/configuration/spec.md#concurrency-scheduler)
 component.
 
 <!-- Configuration Marker -->
@@ -107,7 +107,7 @@
     description='Concurrency Limiter Parameters.'
     type='Object (aperture.spec.v1.ConcurrencyLimiterParameters)'
     reference='../../configuration/spec#concurrency-limiter-parameters'
-    value='{"limit_by_label_key": "limit_by_label_key", "max_inflight_duration": "__REQUIRED_FIELD__"}'
+    value='{"max_inflight_duration": "__REQUIRED_FIELD__"}'
 />
 
 <!-- vale on -->
@@ -135,7 +135,7 @@
     description='Scheduler configuration.'
     type='Object (aperture.spec.v1.Scheduler)'
     reference='../../configuration/spec#scheduler'
-    value='{"priority_label_key": "priority", "tokens_label_key": "tokens", "workload_label_key": "workload"}'
+    value='{}'
 />
 
 <!-- vale on -->

docs/content/reference/blueprints/quota-scheduling/base.md

@@ -8,7 +8,7 @@
 ## Introduction
 
 This blueprint provides a
 [token bucket](https://en.wikipedia.org/wiki/Token_bucket) based quota scheduler
 policy and a dashboard. This policy uses the
 [`QuotaScheduler`](/reference/configuration/spec.md#quota-scheduler) component.
 
@@ -135,7 +135,7 @@
     description='Rate Limiter Parameters.'
     type='Object (aperture.spec.v1.RateLimiterParameters)'
     reference='../../configuration/spec#rate-limiter-parameters'
-    value='{"interval": "__REQUIRED_FIELD__", "limit_by_label_key": "limit_key"}'
+    value='{"interval": "__REQUIRED_FIELD__"}'
 />
 
 <!-- vale on -->
@@ -149,7 +149,7 @@
     description='Scheduler configuration.'
     type='Object (aperture.spec.v1.Scheduler)'
     reference='../../configuration/spec#scheduler'
-    value='{"priority_label_key": "priority", "tokens_label_key": "tokens", "workload_label_key": "workload"}'
+    value='{}'
 />
 
 <!-- vale on -->

docs/content/reference/blueprints/rate-limiting/base.md

@@ -135,7 +135,7 @@ href={`https://github.com/fluxninja/aperture/tree/${aver}/blueprints/rate-limiti
     description='Parameters.'
     type='Object (aperture.spec.v1.RateLimiterParameters)'
     reference='../../configuration/spec#rate-limiter-parameters'
-    value='{"interval": "__REQUIRED_FIELD__", "limit_by_label_key": "limit_key"}'
+    value='{"interval": "__REQUIRED_FIELD__"}'
 />
 
 <!-- vale on -->

pkg/scheduler/wfq.go

@@ -528,11 +528,32 @@ func (pMetrics *preemptionMetrics) onQueueEntry(request *Request, qRequest *queu
 // Update metrics for preemption and delay
 // WARNING: Unsafe and should be called with scheduler lock.
 func (pMetrics *preemptionMetrics) onQueueExit(request *Request, qRequest *queuedRequest, allowed bool) {
+	initMetrics := func(labels prometheus.Labels) error {
+		var err error
+		_, err = pMetrics.workloadPreemptedTokensSummary.GetMetricWith(labels)
+		if err != nil {
+			return fmt.Errorf("%w: failed to get workload_preempted_tokens summary", err)
+		}
+		_, err = pMetrics.workloadDelayedTokensSummary.GetMetricWith(labels)
+		if err != nil {
+			return fmt.Errorf("%w: failed to get workload_delayed_tokens summary", err)
+		}
+		_, err = pMetrics.workloadOnTimeCounter.GetMetricWith(labels)
+		if err != nil {
+			return fmt.Errorf("%w: failed to get workload_on_time_total counter", err)
+		}
+		return nil
+	}
+
 	publishSummary := func(summary *prometheus.SummaryVec, value float64) {
 		if summary == nil {
 			return
 		}
 		metricsLabels := appendWorkloadLabel(pMetrics.metricsLabels, request.FairnessLabel)
+		err := initMetrics(metricsLabels)
+		if err != nil {
+			log.Error().Err(err).Msg("Failed to initialize metrics")
+		}
 		observer, err := summary.GetMetricWith(metricsLabels)
 		if err != nil {
 			log.Error().Err(err).Msg("Failed to get workload preempted tokens summary")
@@ -546,6 +567,10 @@ func (pMetrics *preemptionMetrics) onQueueExit(request *Request, qRequest *queue
 			return
 		}
 		metricsLabels := appendWorkloadLabel(pMetrics.metricsLabels, request.FairnessLabel)
+		err := initMetrics(metricsLabels)
+		if err != nil {
+			log.Error().Err(err).Msg("Failed to initialize metrics")
+		}
 		counter, err := counterVec.GetMetricWith(metricsLabels)
 		if err != nil {
 			log.Error().Err(err).Msg("Failed to get workload on time counter")

playground/scenarios/concurrency-scheduler/policies/concurrency-scheduler-cr.yaml

@@ -20,9 +20,6 @@ spec:
             accept_percentage:
               signal_name: ACCEPT_PERCENTAGE
           scheduler:
-            priority_label_key: priority
-            tokens_label_key: tokens
-            workload_label_key: workload
             workloads:
             - label_matcher:
                 match_labels:

playground/scenarios/postgres-concurrency/load-generator/test.js

@@ -0,0 +1,50 @@
+import { randomIntBetween } from "https://jslib.k6.io/k6-utils/1.4.0/index.js";
+import { check, sleep } from "k6";
+import { vu } from "k6/execution";
+import http from "k6/http";
+
+export let vuStages = [
+  { duration: "10s", target: 10 },
+  { duration: "2m", target: 50 },
+  { duration: "10s", target: 10 },
+];
+
+export let options = {
+  discardResponseBodies: true,
+  scenarios: {
+    guests: {
+      executor: "ramping-vus",
+      stages: vuStages,
+      env: { USER_TYPE: "guest" },
+    },
+    subscribers: {
+      executor: "ramping-vus",
+      stages: vuStages,
+      env: { USER_TYPE: "subscriber" },
+    },
+  },
+};
+
+export default function () {
+  let userType = __ENV.USER_TYPE;
+  let userId = vu.idInTest;
+  const url = "http://localhost:8099/postgres";
+  const headers = {
+    "Content-Type": "application/json",
+    Cookie:
+      "session=eyJ1c2VyIjoia2Vub2JpIn0.YbsY4Q.kTaKRTyOIfVlIbNB48d9YH6Q0wo",
+    "User-Type": userType,
+    "User-Id": userId,
+  };
+  const body = {}
+  let res = http.request("POST", url, JSON.stringify(body), {
+    headers: headers,
+  });
+  const ret = check(res, {
+    "http status was 200": res.status === 200,
+  });
+  if (!ret) {
+    // sleep for 10ms to 25ms
+    sleep(randomIntBetween(0.01, 0.025));
+  }
+}

playground/scenarios/postgres-concurrency/metadata.json

@@ -0,0 +1,26 @@
+{
+  "renderer": "tanka",
+  "tkenv": "playground/tanka/apps/aperture-go-example",
+  "needs": ["postgresql"],
+  "aperture_policies": [
+    {
+      "policy_name": "concurrency-scheduling",
+      "values_file": "policies/concurrency-scheduling.yaml"
+    }
+  ],
+  "images": [
+    {
+      "ref": "aperture-go-example",
+      "context": "sdks/aperture-go",
+      "dockerfile": "Dockerfile.manual",
+      "ssh": "default"
+    }
+  ],
+  "child_resources": [
+    {
+      "workload": "aperture-go-example",
+      "resource_deps": ["cluster-bootstrap", "postgresql"],
+      "extra_objects": ["aperture-go-example:serviceaccount"]
+    }
+  ]
+}