Skip to content

ROSA Cluster - Scaling Benchmark #8

ROSA Cluster - Scaling Benchmark

ROSA Cluster - Scaling Benchmark #8

name: ROSA Cluster - Scaling Benchmark
on:
workflow_dispatch:
inputs:
clusterName:
description: 'Name of the cluster'
type: string
default: 'gh-keycloak'
region:
description: 'Name of the region where EC2 instances should be installed'
type: string
default: 'eu-west-1'
numberOfEntitiesInRealm:
description: 'Number of entities for the scenario in DB'
type: number
default: 100000
maxWaitEntityCreation:
description: 'Maximum number of seconds to wait for creation of entities'
type: number
default: 900
numberOfUsersPerSecond:
description: 'User logins per second'
type: number
default: 200
numberOfClientsPerSecond:
description: 'Client credential grants per second'
type: number
default: 1000
measurement:
description: 'Measurement period (seconds)'
type: number
default: 600
skipCreateDeployment:
description: 'Skip creating Keycloak deployment'
type: boolean
default: false
skipCreateDataset:
description: 'Skip creating dataset'
type: boolean
default: false
skipDeleteProject:
description: 'Skip deleting project'
type: boolean
default: false
concurrency: cluster_${{ github.event.inputs.clusterName || format('gh-{0}', github.repository_owner) }}
env:
PROJECT_PREFIX: runner- # same as default
PROJECT: runner-keycloak
KC_CPU_REQUESTS: 6
KC_INSTANCES: 3
KC_DISABLE_STICKY_SESSION: true
KC_MEMORY_REQUESTS_MB: 4000
KC_MEMORY_LIMITS_MB: 4000
KC_HEAP_MAX_MB: 2048
KC_DB_POOL_INITIAL_SIZE: 30
KC_DB_POOL_MAX_SIZE: 30
KC_DB_POOL_MIN_SIZE: 30
ANSIBLE_CUSTOM_VARS_ARG: '-e @env_rosa_benchmark.yml'
jobs:
run:
name: Run Benchmark
runs-on: ubuntu-latest
steps:
- name: Checkout repository
uses: actions/checkout@v3
- name: Setup ROSA CLI
uses: ./.github/actions/rosa-cli-setup
with:
aws-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
aws-default-region: ${{ vars.AWS_DEFAULT_REGION }}
rosa-token: ${{ secrets.ROSA_TOKEN }}
- name: Login to OpenShift cluster
uses: ./.github/actions/oc-keycloak-login
with:
clusterName: ${{ inputs.clusterName || format('gh-{0}', github.repository_owner) }}
- name: Set up JDK
uses: actions/setup-java@v3
with:
distribution: 'temurin'
java-version: '17'
cache: 'maven'
- name: Cache Maven Wrapper
uses: actions/cache@v3
with:
path: |
.mvn/wrapper/maven-wrapper.jar
key: ${{ runner.os }}-maven-wrapper-${{ hashFiles('**/maven-wrapper.properties') }}
restore-keys: |
${{ runner.os }}-maven-wrapper-
- name: Build with Maven
run: |
./mvnw -B clean package -DskipTests -pl benchmark
tar xfvz benchmark/target/keycloak-benchmark-*.tar.gz
mv keycloak-benchmark-* keycloak-benchmark
- name: Allow cluster to scale
if: ${{ !inputs.skipCreateDeployment }}
run: rosa edit machinepool -c ${{ inputs.clusterName }} --min-replicas 3 --max-replicas 10 scaling
- name: Create Keycloak deployment
if: ${{ !inputs.skipCreateDeployment }}
uses: ./.github/actions/keycloak-create-deployment
with:
projectPrefix: ${{ env.PROJECT_PREFIX }}
disableStickySessions: ${{ !env.KC_DISABLE_STICKY_SESSION }}
replicas: ${{ env.KC_INSTANCES }}
podCpuRequests: ${{ env.KC_CPU_REQUESTS }}
podMemoryRequests: ${{ env.KC_MEMORY_REQUESTS_MB }}
podMemoryLimit: ${{ env.KC_MEMORY_LIMITS_MB }}
heapMaxSizeMB: ${{ env.KC_HEAP_MAX_MB }}
- name: Get URLs
uses: ./.github/actions/get-keycloak-url
with:
project: ${{ env.PROJECT }}
- name: Create Keycloak dataset with "${{ inputs.numberOfEntitiesInRealm }}" users and clients
if: ${{ !inputs.skipCreateDataset }}
uses: ./.github/actions/keycloak-create-dataset
with:
project: ${{ env.PROJECT }}
users: ${{ inputs.numberOfEntitiesInRealm }}
clients: 100
clientsPerRealm: ${{ inputs.numberOfEntitiesInRealm }}
createClientForSpecificRealm: true
maxWaitEntityCreation: ${{ inputs.maxWaitEntityCreation }}
- name: Create AWS EC2 instances
id: create_aws_ec2_instances
uses: ./.github/actions/ec2-create-instances
with:
region: ${{ inputs.region }}
- name: Run Memory Usage Total Query Before Benchmark
uses: ./.github/actions/prometheus-run-queries
with:
project: ${{ env.PROJECT }}
runMemoryUsageTotal: true
output: memory_create_sessions
- name: Testing memory for creating sessions
id: kcb-authorization-code-1
run: >
./benchmark.sh ${{ inputs.region }}
--scenario=keycloak.scenario.authentication.AuthorizationCode
--server-url=${{ env.KEYCLOAK_URL }}
--realm-name=realm-0
--users-per-sec=${{ inputs.numberOfUsersPerSecond }}
--ramp-up=20
--logout-percentage=0
--measurement=${{ inputs.measurement }}
--users-per-realm=${{ inputs.numberOfEntitiesInRealm }}
--log-http-on-failure
--sla-error-percentage=0.001
continue-on-error: true
working-directory: ansible
- name: Run Memory Usage Total Query After Benchmark
uses: ./.github/actions/prometheus-run-queries
with:
project: ${{ env.PROJECT }}
runMemoryUsageTotal: true
output: memory_create_sessions
- name: Calculate number of active sessions
id: active_sessions_count
env:
USERS_PER_SEC: ${{ inputs.numberOfUsersPerSecond }}
TIME_IN_SEC: ${{ inputs.measurement }}
run: echo "active_sessions=$(awk "BEGIN {print $USERS_PER_SEC*$TIME_IN_SEC; exit}")" >> $GITHUB_OUTPUT
- name: Calculate and Store Memory Usage Total For Active User Sessions
uses: ./.github/actions/prometheus-metrics-calc
with:
input: memory_create_sessions
performedTestName: 'Memory usage for sessions'
calculatedMetricName: 'Active sessions per 500 MB memory' #as a unit for memory calculation is chosen 500MB memory size to find out how much user sessions so much memory can handle.
criteriaValue: ${{ steps.active_sessions_count.outputs.active_sessions }}
isvCPU: false
isMemory: true
- name: Run CPU sec Util Query Before Benchmark
uses: ./.github/actions/prometheus-run-queries
with:
project: ${{ env.PROJECT }}
runCpuSecsUtil: true
output: user_logins_vCpu
- name: Testing CPU usage for user logins
id: kcb-authorization-code-2
run: >
./benchmark.sh ${{ inputs.region }}
--scenario=keycloak.scenario.authentication.AuthorizationCode
--server-url=${{ env.KEYCLOAK_URL }}
--realm-name=realm-0
--users-per-sec=${{ inputs.numberOfUsersPerSecond }}
--ramp-up=20
--logout-percentage=100
--measurement=${{ inputs.measurement }}
--users-per-realm=${{ inputs.numberOfEntitiesInRealm }}
--log-http-on-failure
--sla-error-percentage=0.001
continue-on-error: true
working-directory: ansible
- name: Run CPU sec Util Query After Benchmark
uses: ./.github/actions/prometheus-run-queries
with:
project: ${{ env.PROJECT }}
runCpuSecsUtil: true
output: user_logins_vCpu
- name: Calculate and Store CPU sec Util For Ran Benchmark
uses: ./.github/actions/prometheus-metrics-calc
with:
input: user_logins_vCpu
performedTestName: 'CPU usage for user logins'
calculatedMetricName: 'User Logins per second per 1vCPU'
criteriaValue: ${{ inputs.numberOfUsersPerSecond }}
measurementInterval: ${{ inputs.measurement }}
isvCPU: true
isMemory: false
- name: Run CPU sec Util Query Before Benchmark
uses: ./.github/actions/prometheus-run-queries
with:
project: ${{ env.PROJECT }}
runCpuSecsUtil: true
output: client_credential_grants_vCpu
- name: Testing CPU usage for client credential grants
id: kcb-client-secret
run: >
./benchmark.sh ${{ inputs.region }}
--scenario=keycloak.scenario.authentication.ClientSecret
--server-url=${{ env.KEYCLOAK_URL }}
--realm-name=realm-0
--users-per-sec=${{ inputs.numberOfClientsPerSecond }}
--ramp-up=20
--measurement=${{ inputs.measurement }}
--users-per-realm=${{ inputs.numberOfEntitiesInRealm }}
--log-http-on-failure
--sla-error-percentage=0.001
continue-on-error: true
working-directory: ansible
- name: Run CPU sec Util Query After Benchmark
uses: ./.github/actions/prometheus-run-queries
with:
project: ${{ env.PROJECT }}
runCpuSecsUtil: true
output: client_credential_grants_vCpu
- name: Calculate and Store CPU usage For Ran Benchmark
uses: ./.github/actions/prometheus-metrics-calc
with:
input: client_credential_grants_vCpu
performedTestName: 'CPU usage for client credential grants'
calculatedMetricName: 'Client Credential Grants per second per 1vCPU'
criteriaValue: ${{ inputs.numberOfClientsPerSecond }}
measurementInterval: ${{ inputs.measurement }}
isvCPU: true
isMemory: false
- name: Archive Gatling reports
if: ${{ always() }}
uses: actions/upload-artifact@v3
with:
name: gatling-results
path: ansible/files/benchmark/*/results
retention-days: 5
- name: Archive Summary
uses: actions/upload-artifact@v3
with:
name: summary
path: ansible/files/benchmark/*/results/*/js/stats.json
retention-days: 5
- name: Archive Calculated Metrics summary
uses: actions/upload-artifact@v3
with:
name: Calculated environment conclusion
path: |
result-*.json
retention-days: 5
- name: Delete EC2 instances
if: ${{ always() && steps.create_aws_ec2_instances.conclusion != 'skipped' }}
uses: ./.github/actions/ec2-delete-instances
with:
region: ${{ inputs.region }}
- name: Delete Keycloak deployment
if: ${{ (success() || failure()) && !inputs.skipDeleteProject }}
uses: ./.github/actions/keycloak-delete-deployment
with:
project: ${{ env.PROJECT }}
- name: Scale down the cluster
if: ${{ (success() || failure()) && !inputs.skipDeleteProject }}
run: rosa edit machinepool -c ${{ inputs.clusterName }} --min-replicas 0 --max-replicas 0 scaling
archive:
name: Commit results to Git repository
runs-on: ubuntu-latest
permissions:
contents: write
actions: write
needs:
- run
steps:
- name: Checkout repository for results
uses: actions/checkout@v3
with:
ref: 'result-data'
- uses: actions/download-artifact@v3
with:
name: Calculated environment conclusion
- name: Commit result-summary
shell: bash
env:
GITHUB_OAUTH: ${{ secrets.GITHUB_TOKEN }}
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
git config --global user.email "41898282+github-actions[bot]@users.noreply.github.com"
git config --global user.name "github-actions[bot]"
DATE_FOLDER=rosa_scalability/$(date +"%Y/%m/%d")
mkdir -p ${DATE_FOLDER}
mv *.json ${DATE_FOLDER}
git add .
git commit -m "generated"
git push
- name: Trigger data aggregation
if: github.repository == 'keycloak/keycloak-benchmark'
env:
GH_TOKEN: ${{ github.token }}
# manually trigger the run, as a push with a standard GitHub action token doesn't trigger any workflow run on GitHub
run: gh workflow run -R keycloak/keycloak-benchmark aggregate-results.yaml --ref result-data