Add pipeline for konnectivity agent autoscaler performance

Signed-off-by: Kartik Joshi <[email protected]>
Azure · Jan 20, 2025 · d7c0a7f · d7c0a7f
1 parent 0ad5629
commit d7c0a7f
Show file tree

Hide file tree

Showing 8 changed files with 199 additions and 0 deletions.
diff --git a/pipelines/perf-eval/CRI Autoscale Benchmark/cri-autoscale.yml b/pipelines/perf-eval/CRI Autoscale Benchmark/cri-autoscale.yml
@@ -0,0 +1,40 @@
+trigger: none
+  #schedules:
+  #- cron: "0 */12 * * *"
+  #  displayName: "Every 12 Hours"
+  #  branches:
+  #    include:
+  #      - main
+  #  always: true
+
+variables:
+  SCENARIO_TYPE: perf-eval
+  SCENARIO_NAME: konnectivity-autoscale
+  SCENARIO_VERSION: main
+
+stages:
+  - stage: azure_eastus
+    dependsOn: []
+    jobs:
+      - template: /jobs/competitive-test.yml
+        parameters:
+          cloud: azure
+          regions:
+            - westeurope
+          engine: clusterloader2
+          engine_input:
+            image: "ghcr.io/azure/clusterloader2:v20241016"
+          topology: cri-autoscale
+          matrix:
+            n10:
+              node_count: 10
+              scale_up_timeout: "10m"
+              scale_down_timeout: "10m"
+            n100:
+              node_count: 100
+              scale_up_timeout: "15m"
+              scale_down_timeout: "15m"
+          max_parallel: 2
+          timeout_in_minutes: 240
+          credential_type: service_connection
+          ssh_key_enabled: false
diff --git a/scenarios/perf-eval/konnectivity-autoscale/terraform-inputs/azure.tfvars b/scenarios/perf-eval/konnectivity-autoscale/terraform-inputs/azure.tfvars
@@ -0,0 +1,62 @@
+scenario_type  = "perf-eval"
+scenario_name  = "konnectivity-autoscale"
+deletion_delay = "240h"
+owner          = "aks"
+
+network_config_list = [
+  {
+    role               = "konnectivity"
+    vnet_name          = "cri-autoscale-vnet"
+    vnet_address_space = "10.0.0.0/9"
+    subnet = [
+      {
+        name           = "cri-autoscale-subnet-1"
+        address_prefix = "10.0.0.0/22"
+      }
+    ]
+    network_security_group_name = ""
+    nic_public_ip_associations  = []
+    nsr_rules                   = []
+  }
+]
+
+aks_config_list = [
+  {
+    role        = "konnectivity"
+    aks_name    = "konnectivity-autoscale"
+    dns_prefix  = "cl2"
+    subnet_name = "aks-network"
+    sku_tier    = "Standard"
+    network_profile = {
+      network_plugin      = "azure"
+      network_plugin_mode = "overlay"
+    }
+    default_node_pool = {
+      name                         = "default"
+      node_count                   = 3
+      vm_size                      = "Standard_D16s_v3"
+      os_disk_type                 = "Managed"
+      only_critical_addons_enabled = true
+      temporary_name_for_rotation  = "defaulttmp"
+    }
+    extra_node_pool = [
+      {
+        name                 = "prompool"
+        node_count           = 1
+        auto_scaling_enabled = false
+        vm_size              = "Standard_D16_v3"
+        node_labels          = { "prometheus" = "true" }
+      },
+      {
+        name                 = "userpool"
+        node_count           = 1
+        min_count            = 0
+        max_count            = 1000
+        auto_scaling_enabled = true
+        vm_size              = "Standard_D16s_v3"
+        max_pods             = 110
+      }
+    ]
+    kubernetes_version = "1.30"
+  }
+]
diff --git a/scenarios/perf-eval/konnectivity-autoscale/terraform-test-inputs/azure.json b/scenarios/perf-eval/konnectivity-autoscale/terraform-test-inputs/azure.json
@@ -0,0 +1,4 @@
+{
+  "run_id": "1234567890",
+  "region": "eastus",
+}
diff --git a/steps/engine/clusterloader2/cri-autoscale/collect.yml b/steps/engine/clusterloader2/cri-autoscale/collect.yml
@@ -0,0 +1,25 @@
+parameters:
+- name: cloud
+  type: string
+  default: ''
+- name: engine_input
+  type: object
+  default: {}
+- name: region
+  type: string
+
+steps:
+- template: /steps/cloud/${{ parameters.cloud }}/collect-cloud-info.yml
+  parameters:
+    region: ${{ parameters.region }}
+- script: |
+    set -eo pipefail
+
+    PYTHONPATH=$PYTHONPATH:$(pwd) python3 $PYTHON_SCRIPT_FILE collect \
+      $NODE_COUNT $CL2_REPORT_DIR $RESULT_FILE
+  workingDirectory: modules/python/clusterloader2
+  env:
+    CLOUD: ${{ parameters.cloud }}
+    PYTHON_SCRIPT_FILE: $(Pipeline.Workspace)/s/modules/python/clusterloader2/cri-autoscale/cri-autoscale.py
+    CL2_REPORT_DIR: $(Pipeline.Workspace)/s/modules/python/clusterloader2/cri-autoscale/results
+  displayName: "Collect Node Count, Agent Pod Count, and CPU Usage"
diff --git a/steps/engine/clusterloader2/cri-autoscale/execute.yml b/steps/engine/clusterloader2/cri-autoscale/execute.yml
@@ -0,0 +1,29 @@
+parameters:
+- name: cloud
+  type: string
+  default: ''
+- name: engine_input
+  type: object
+  default: {}
+- name: region
+  type: string
+
+steps:
+- script: |
+    set -eo pipefail
+
+    # Override the configuration for scaling
+    PYTHONPATH=$PYTHONPATH:$(pwd) python3 $PYTHON_SCRIPT_FILE override \
+      $NODE_COUNT $SCALE_UP_TIMEOUT $SCALE_DOWN_TIMEOUT ${CL2_CONFIG_DIR}/overrides.yaml
+
+    # Execute the scaling workload
+    PYTHONPATH=$PYTHONPATH:$(pwd) python3 $PYTHON_SCRIPT_FILE execute \
+      ${CL2_IMAGE} ${CL2_CONFIG_DIR} $CL2_REPORT_DIR ${HOME}/.kube/config $CLOUD
+  workingDirectory: modules/python/clusterloader2
+  env:
+    CLOUD: ${{ parameters.cloud }}
+    PYTHON_SCRIPT_FILE: $(Pipeline.Workspace)/s/modules/python/clusterloader2/cri-autoscale/cri-autoscale.py
+    CL2_IMAGE: ${{ parameters.engine_input.image }}
+    CL2_CONFIG_DIR: $(Pipeline.Workspace)/s/modules/python/clusterloader2/cri-autoscale/config
+    CL2_REPORT_DIR: $(Pipeline.Workspace)/s/modules/python/clusterloader2/cri-autoscale/results
+  displayName: "Execute Konnectivity Autoscaler Scaling Test"
diff --git a/steps/topology/cri-autoscale/collect-clusterloader2.yml b/steps/topology/cri-autoscale/collect-clusterloader2.yml
@@ -0,0 +1,13 @@
+parameters:
+- name: cloud
+  type: string
+  default: ''
+- name: regions
+  type: object
+  default: []
+
+steps:
+- template: /steps/engine/clusterloader2/cri-autoscale/collect.yml
+  parameters:
+    cloud: ${{ parameters.cloud }}
+    region: ${{ parameters.regions[0] }}
diff --git a/steps/topology/cri-autoscale/execute-clusterloader2.yml b/steps/topology/cri-autoscale/execute-clusterloader2.yml
@@ -0,0 +1,13 @@
+parameters:
+- name: cloud
+  type: string
+  default: ''
+- name: regions
+  type: object
+  default: []
+
+steps:
+- template: /steps/engine/clusterloader2/cri-autoscale/execute.yml
+  parameters:
+    cloud: ${{ parameters.cloud }}
+    region: ${{ parameters.regions[0] }}
diff --git a/steps/topology/cri-autoscale/validate-resources.yml b/steps/topology/cri-autoscale/validate-resources.yml
@@ -0,0 +1,13 @@
+parameters:
+- name: cloud
+  type: string
+- name: engine
+  type: string
+- name: regions
+  type: object
+
+steps:
+- template: /steps/cloud/${{ parameters.cloud }}/update-kubeconfig.yml
+  parameters:
+    role: konnectivity
+    region: ${{ parameters.regions[0] }}