From 8b97cd54218f82cd9dd44e7f35b782525870dae7 Mon Sep 17 00:00:00 2001 From: Rishabh Singh Date: Thu, 18 Jul 2024 11:36:03 -0700 Subject: [PATCH] Run performance benchmark on pull requests (#14760) * add performance benchmark workflow for pull requests Signed-off-by: Rishabh Singh * Update PERFORMANCE_BENCHMARKS.md Co-authored-by: Andriy Redko Signed-off-by: Rishabh Singh * Update PERFORMANCE_BENCHMARKS.md Co-authored-by: Andriy Redko Signed-off-by: Rishabh Singh * Update .github/workflows/benchmark-pull-request.yml Co-authored-by: Andriy Redko Signed-off-by: Rishabh Singh * Update .github/workflows/benchmark-pull-request.yml Co-authored-by: Andriy Redko Signed-off-by: Rishabh Singh * Update .github/workflows/benchmark-pull-request.yml Co-authored-by: Andriy Redko Signed-off-by: Rishabh Singh * Update .github/workflows/benchmark-pull-request.yml Co-authored-by: Andriy Redko Signed-off-by: Rishabh Singh --------- Signed-off-by: Rishabh Singh Signed-off-by: Rishabh Singh Co-authored-by: Andriy Redko Signed-off-by: Kaushal Kumar --- .github/benchmark-configs.json | 155 ++++++++++++++++++ .github/workflows/add-performance-comment.yml | 25 +++ .github/workflows/benchmark-pull-request.yml | 142 ++++++++++++++++ PERFORMANCE_BENCHMARKS.md | 112 +++++++++++++ 4 files changed, 434 insertions(+) create mode 100644 .github/benchmark-configs.json create mode 100644 .github/workflows/add-performance-comment.yml create mode 100644 .github/workflows/benchmark-pull-request.yml create mode 100644 PERFORMANCE_BENCHMARKS.md diff --git a/.github/benchmark-configs.json b/.github/benchmark-configs.json new file mode 100644 index 0000000000000..a5b1951d2240c --- /dev/null +++ b/.github/benchmark-configs.json @@ -0,0 +1,155 @@ +{ + "name": "Cluster and opensearch-benchmark configurations", + "id_1": { + "description": "Indexing only configuration for NYC_TAXIS workload", + "supported_major_versions": ["2", "3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "nyc_taxis", + "WORKLOAD_PARAMS": "{\"number_of_replicas\":\"0\",\"number_of_shards\":\"1\"}", + "EXCLUDE_TASKS": "type:search", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_2": { + "description": "Indexing only configuration for HTTP_LOGS workload", + "supported_major_versions": ["2", "3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "http_logs", + "WORKLOAD_PARAMS": "{\"number_of_replicas\":\"0\",\"number_of_shards\":\"1\"}", + "EXCLUDE_TASKS": "type:search", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_3": { + "description": "Search only test-procedure for NYC_TAXIS, uses snapshot to restore the data for OS-3.0.0", + "supported_major_versions": ["3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "nyc_taxis", + "WORKLOAD_PARAMS": "{\"snapshot_repo_name\":\"benchmark-workloads-repo-300\",\"snapshot_bucket_name\":\"benchmark-workload-snapshots\",\"snapshot_region\":\"us-east-1\",\"snapshot_base_path\":\"workload-snapshots-300\",\"snapshot_name\":\"nyc_taxis_1_shard\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_4": { + "description": "Search only test-procedure for HTTP_LOGS, uses snapshot to restore the data for OS-3.0.0", + "supported_major_versions": ["3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "http_logs", + "WORKLOAD_PARAMS": "{\"snapshot_repo_name\":\"benchmark-workloads-repo-300\",\"snapshot_bucket_name\":\"benchmark-workload-snapshots\",\"snapshot_region\":\"us-east-1\",\"snapshot_base_path\":\"workload-snapshots-300\",\"snapshot_name\":\"http_logs_1_shard\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_5": { + "description": "Search only test-procedure for HTTP_LOGS, uses snapshot to restore the data for OS-3.0.0", + "supported_major_versions": ["3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "big5", + "WORKLOAD_PARAMS": "{\"snapshot_repo_name\":\"benchmark-workloads-repo-300\",\"snapshot_bucket_name\":\"benchmark-workload-snapshots\",\"snapshot_region\":\"us-east-1\",\"snapshot_base_path\":\"workload-snapshots-300\",\"snapshot_name\":\"big5_1_shard\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_6": { + "description": "Search only test-procedure for NYC_TAXIS, uses snapshot to restore the data for OS-2.x", + "supported_major_versions": ["2"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "nyc_taxis", + "WORKLOAD_PARAMS": "{\"snapshot_repo_name\":\"benchmark-workloads-repo\",\"snapshot_bucket_name\":\"benchmark-workload-snapshots\",\"snapshot_region\":\"us-east-1\",\"snapshot_base_path\":\"workload-snapshots\",\"snapshot_name\":\"nyc_taxis_1_shard\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_7": { + "description": "Search only test-procedure for HTTP_LOGS, uses snapshot to restore the data for OS-2.x", + "supported_major_versions": ["2"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "http_logs", + "WORKLOAD_PARAMS": "{\"snapshot_repo_name\":\"benchmark-workloads-repo\",\"snapshot_bucket_name\":\"benchmark-workload-snapshots\",\"snapshot_region\":\"us-east-1\",\"snapshot_base_path\":\"workload-snapshots\",\"snapshot_name\":\"http_logs_1_shard\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_8": { + "description": "Search only test-procedure for HTTP_LOGS, uses snapshot to restore the data for OS-2.x", + "supported_major_versions": ["2"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "big5", + "WORKLOAD_PARAMS": "{\"snapshot_repo_name\":\"benchmark-workloads-repo\",\"snapshot_bucket_name\":\"benchmark-workload-snapshots\",\"snapshot_region\":\"us-east-1\",\"snapshot_base_path\":\"workload-snapshots\",\"snapshot_name\":\"big5_1_shard\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_9": { + "description": "Indexing and search configuration for pmc workload", + "supported_major_versions": ["2", "3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "pmc", + "WORKLOAD_PARAMS": "{\"number_of_replicas\":\"0\",\"number_of_shards\":\"1\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + }, + "id_10": { + "description": "Indexing only configuration for stack-overflow workload", + "supported_major_versions": ["2", "3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "true", + "MIN_DISTRIBUTION": "true", + "TEST_WORKLOAD": "so", + "WORKLOAD_PARAMS": "{\"number_of_replicas\":\"0\",\"number_of_shards\":\"1\"}", + "CAPTURE_NODE_STAT": "true" + }, + "cluster_configuration": { + "size": "Single-Node", + "data_instance_config": "4vCPU, 32G Mem, 16G Heap" + } + } +} diff --git a/.github/workflows/add-performance-comment.yml b/.github/workflows/add-performance-comment.yml new file mode 100644 index 0000000000000..3939de25e4cbe --- /dev/null +++ b/.github/workflows/add-performance-comment.yml @@ -0,0 +1,25 @@ +name: Performance Label Action + +on: + pull_request: + types: [labeled] + +jobs: + add-comment: + if: github.event.label.name == 'Performance' + runs-on: ubuntu-latest + permissions: + pull-requests: write + + steps: + - name: Add comment to PR + uses: actions/github-script@v6 + with: + github-token: ${{secrets.GITHUB_TOKEN}} + script: | + github.rest.issues.createComment({ + issue_number: context.issue.number, + owner: context.repo.owner, + repo: context.repo.repo, + body: "Hello!\nWe have added a performance benchmark workflow that runs by adding a comment on the PR.\n Please refer https://github.com/opensearch-project/OpenSearch/blob/main/PERFORMANCE_BENCHMARKS.md on how to run benchmarks on pull requests." + }) diff --git a/.github/workflows/benchmark-pull-request.yml b/.github/workflows/benchmark-pull-request.yml new file mode 100644 index 0000000000000..0de50981fa3d7 --- /dev/null +++ b/.github/workflows/benchmark-pull-request.yml @@ -0,0 +1,142 @@ +name: Run performance benchmark on pull request +on: + issue_comment: + types: [created] +jobs: + run-performance-benchmark-on-pull-request: + if: ${{ (github.event.issue.pull_request) && (contains(github.event.comment.body, '"run-benchmark-test"')) }} + runs-on: ubuntu-latest + permissions: + id-token: write + contents: read + issues: write + pull-requests: write + steps: + - name: Checkout Repository + uses: actions/checkout@v3 + - name: Set up required env vars + run: | + echo "PR_NUMBER=${{ github.event.issue.number }}" >> $GITHUB_ENV + echo "REPOSITORY=${{ github.event.repository.full_name }}" >> $GITHUB_ENV + OPENSEARCH_VERSION=$(awk -F '=' '/^opensearch[[:space:]]*=/ {gsub(/[[:space:]]/, "", $2); print $2}' buildSrc/version.properties) + echo "OPENSEARCH_VERSION=$OPENSEARCH_VERSION" >> $GITHUB_ENV + major_version=$(echo $OPENSEARCH_VERSION | cut -d'.' -f1) + echo "OPENSEARCH_MAJOR_VERSION=$major_version" >> $GITHUB_ENV + echo "USER_TAGS=pull_request_number:${{ github.event.issue.number }},repository:OpenSearch" >> $GITHUB_ENV + - name: Check comment format + id: check_comment + run: | + comment='${{ github.event.comment.body }}' + if echo "$comment" | jq -e 'has("run-benchmark-test")'; then + echo "Valid comment format detected, check if valid config id is provided" + config_id=$(echo $comment | jq -r '.["run-benchmark-test"]') + benchmark_configs=$(cat .github/benchmark-configs.json) + if echo $benchmark_configs | jq -e --arg id "$config_id" 'has($id)' && echo "$benchmark_configs" | jq -e --arg version "$OPENSEARCH_MAJOR_VERSION" --arg id "$config_id" '.[$id].supported_major_versions | index($version) != null' > /dev/null; then + echo $benchmark_configs | jq -r --arg id "$config_id" '.[$id]."cluster-benchmark-configs" | to_entries[] | "\(.key)=\(.value)"' >> $GITHUB_ENV + else + echo "invalid=true" >> $GITHUB_OUTPUT + fi + else + echo "invalid=true" >> $GITHUB_OUTPUT + fi + - name: Post invalid format comment + if: steps.check_comment.outputs.invalid == 'true' + uses: actions/github-script@v6 + with: + github-token: ${{secrets.GITHUB_TOKEN}} + script: | + github.rest.issues.createComment({ + issue_number: context.issue.number, + owner: context.repo.owner, + repo: context.repo.repo, + body: 'Invalid comment format or config id. Please refer to https://github.com/opensearch-project/OpenSearch/blob/main/PERFORMANCE_BENCHMARKS.md on how to run benchmarks on pull requests.' + }) + - name: Fail workflow for invalid comment + if: steps.check_comment.outputs.invalid == 'true' + run: | + echo "Invalid comment format detected. Failing the workflow." + exit 1 + - id: get_approvers + run: | + echo "approvers=$(cat .github/CODEOWNERS | grep '^\*' | tr -d '* ' | sed 's/@/,/g' | sed 's/,//1')" >> $GITHUB_OUTPUT + - uses: trstringer/manual-approval@v1 + if: (!contains(steps.get_approvers.outputs.approvers, github.event.comment.user.login)) + with: + secret: ${{ github.TOKEN }} + approvers: ${{ steps.get_approvers.outputs.approvers }} + minimum-approvals: 1 + issue-title: 'Request to approve/deny benchmark run for PR #${{ env.PR_NUMBER }}' + issue-body: "Please approve or deny the benchmark run for PR #${{ env.PR_NUMBER }}" + exclude-workflow-initiator-as-approver: false + - name: Get PR Details + id: get_pr + uses: actions/github-script@v7 + with: + script: | + const issue = context.payload.issue; + const prNumber = issue.number; + console.log(`Pull Request Number: ${prNumber}`); + + const { data: pull_request } = await github.rest.pulls.get({ + owner: context.repo.owner, + repo: context.repo.repo, + pull_number: prNumber, + }); + + return { + "headRepoFullName": pull_request.head.repo.full_name, + "headRef": pull_request.head.ref + }; + - name: Set pr details env vars + run: | + echo '${{ steps.get_pr.outputs.result }}' | jq -r '.headRepoFullName' + echo '${{ steps.get_pr.outputs.result }}' | jq -r '.headRef' + headRepo=$(echo '${{ steps.get_pr.outputs.result }}' | jq -r '.headRepoFullName') + headRef=$(echo '${{ steps.get_pr.outputs.result }}' | jq -r '.headRef') + echo "prHeadRepo=$headRepo" >> $GITHUB_ENV + echo "prheadRef=$headRef" >> $GITHUB_ENV + - name: Checkout PR Repo + uses: actions/checkout@v2 + with: + repository: ${{ env.prHeadRepo }} + ref: ${{ env.prHeadRef }} + token: ${{ secrets.GITHUB_TOKEN }} + - name: Setup Java + uses: actions/setup-java@v1 + with: + java-version: 21 + - name: Build and Assemble OpenSearch from PR + run: | + ./gradlew :distribution:archives:linux-tar:assemble -Dbuild.snapshot=false + - name: Configure AWS credentials + uses: aws-actions/configure-aws-credentials@v4 + with: + role-to-assume: ${{ secrets.UPLOAD_ARCHIVE_ARTIFACT_ROLE }} + role-session-name: publish-to-s3 + aws-region: us-west-2 + - name: Push to S3 + run: | + aws s3 cp distribution/archives/linux-tar/build/distributions/opensearch-min-$OPENSEARCH_VERSION-linux-x64.tar.gz s3://${{ secrets.ARCHIVE_ARTIFACT_BUCKET_NAME }}/PR-$PR_NUMBER/ + echo "DISTRIBUTION_URL=${{ secrets.ARTIFACT_BUCKET_CLOUDFRONT_URL }}/PR-$PR_NUMBER/opensearch-min-$OPENSEARCH_VERSION-linux-x64.tar.gz" >> $GITHUB_ENV + - name: Checkout opensearch-build repo + uses: actions/checkout@v4 + with: + repository: opensearch-project/opensearch-build + ref: main + path: opensearch-build + - name: Trigger jenkins workflow to run gradle check + run: | + cat $GITHUB_ENV + bash opensearch-build/scripts/benchmark/benchmark-pull-request.sh ${{ secrets.JENKINS_PR_BENCHMARK_GENERIC_WEBHOOK_TOKEN }} + - name: Update PR with Job Url + uses: actions/github-script@v6 + with: + github-token: ${{ secrets.GITHUB_TOKEN }} + script: | + const workflowUrl = process.env.WORKFLOW_URL; + github.rest.issues.createComment({ + issue_number: context.issue.number, + owner: context.repo.owner, + repo: context.repo.repo, + body: `The Jenkins job url is ${workflowUrl} . Final results will be published once the job is completed.` + }) diff --git a/PERFORMANCE_BENCHMARKS.md b/PERFORMANCE_BENCHMARKS.md new file mode 100644 index 0000000000000..252c4ae312136 --- /dev/null +++ b/PERFORMANCE_BENCHMARKS.md @@ -0,0 +1,112 @@ +# README: Running Performance Benchmarks on Pull Requests + +## Overview + +`benchmark-pull-request` GitHub Actions workflow is designed to automatically run performance benchmarks on a pull request when a specific comment is made on the pull request. This ensures that performance benchmarks are consistently and accurately applied to code changes, helping maintain the performance standards of the repository. + +## Workflow Trigger + +The workflow is triggered when a new comment is created on a pull request. Specifically, it checks for the presence of the `"run-benchmark-test"` keyword in the comment body. If this keyword is detected, the workflow proceeds to run the performance benchmarks. + +## Key Steps in the Workflow + +1. **Check Comment Format and Configuration:** + - Validates the format of the comment to ensure it contains the required `"run-benchmark-test"` keyword and is in json format. + - Extracts the benchmark configuration ID from the comment and verifies if it exists in the `benchmark-config.json` file. + - Checks if the extracted configuration ID is supported for the current OpenSearch major version. + +2. **Post Invalid Format Comment:** + - If the comment format is invalid or the configuration ID is not supported, a comment is posted on the pull request indicating the problem, and the workflow fails. + +3. **Manual Approval (if necessary):** + - Fetches the list of approvers from the `.github/CODEOWNERS` file. + - If the commenter is not one of the maintainers, a manual approval request is created. The workflow pauses until an approver approves or denies the benchmark run by commenting appropriate word on the issue. + - The issue for approval request is auto-closed once the approver is done adding appropriate comment + +4. **Build and Assemble OpenSearch:** + - Builds and assembles (x64-linux tar) the OpenSearch distribution from the pull request code changes. + +5. **Upload to S3:** + - Configures AWS credentials and uploads the assembled OpenSearch distribution to an S3 bucket for further use in benchmarking. + - The S3 bucket is fronted by cloudfront to only allow downloads. + - The lifecycle policy on the S3 bucket will delete the uploaded artifacts after 30-days. + +6. **Trigger Jenkins Workflow:** + - Triggers a Jenkins workflow to run the benchmark tests using a webhook token. + +7. **Update Pull Request with Job URL:** + - Posts a comment on the pull request with the URL of the Jenkins job. The final benchmark results will be posted once the job completes. + - To learn about how benchmark job works see https://github.com/opensearch-project/opensearch-build/tree/main/src/test_workflow#benchmarking-tests + +## How to Use This Workflow + +1. **Ensure `benchmark-config.json` is Up-to-Date:** + - The `benchmark-config.json` file should contain valid benchmark configurations with supported major versions and cluster-benchmark configurations. + +2. **Add the Workflow to Your Repository:** + - Save the workflow YAML file (typically named `benchmark.yml`) in the `.github/workflows` directory of your repository. + +3. **Make a Comment to Trigger the Workflow:** + - On any pull request issue, make a comment containing the keyword `"run-benchmark-test"` along with the configuration ID. For example: + ```json + {"run-benchmark-test": "id_1"} + ``` + +4. **Monitor Workflow Progress:** + - The workflow will validate the comment, check for approval (if necessary), build the OpenSearch distribution, and trigger the Jenkins job. + - A comment will be posted on the pull request with the URL of the Jenkins job. You can monitor the progress and final results there as well. + +## Example Comment Format + +To run the benchmark with configuration ID `id_1`, post the following comment on the pull request issue: +```json +{"run-benchmark-test": "id_1"} +``` + +## How to add a new benchmark configuration + +The benchmark-config.json file accepts the following schema. +```json +{ + "id_": { + "description": "Short description of the configuration", + "supported_major_versions": ["2", "3"], + "cluster-benchmark-configs": { + "SINGLE_NODE_CLUSTER": "Use single node cluster for benchmarking, accepted values are \"true\" or \"false\"", + "MIN_DISTRIBUTION": "Use OpenSearch min distribution, should always be \"true\"", + "MANAGER_NODE_COUNT": "For multi-node cluster tests, number of cluster manager nodes, empty value defaults to 3.", + "DATA_NODE_COUNT": "For multi-node cluster tests, number of data nodes, empty value defaults to 2.", + "DATA_INSTANCE_TYPE": "EC2 instance type for data node, empty defaults to r5.xlarge.", + "DATA_NODE_STORAGE": "Data node ebs block storage size, empty value defaults to 100Gb", + "JVM_SYS_PROPS": "A comma-separated list of key=value pairs that will be added to jvm.options as JVM system properties", + "ADDITIONAL_CONFIG": "Additional space delimited opensearch.yml config parameters. e.g., `search.concurrent_segment_search.enabled:true`", + "TEST_WORKLOAD": "The workload name from OpenSearch Benchmark Workloads. https://github.com/opensearch-project/opensearch-benchmark-workloads. Default is nyc_taxis", + "WORKLOAD_PARAMS": "With this parameter you can inject variables into workloads, e.g.{\"number_of_replicas\":\"0\",\"number_of_shards\":\"3\"}. See https://opensearch.org/docs/latest/benchmark/reference/commands/command-flags/#workload-params", + "EXCLUDE_TASKS": "Defines a comma-separated list of test procedure tasks not to run. e.g. type:search, see https://opensearch.org/docs/latest/benchmark/reference/commands/command-flags/#exclude-tasks", + "INCLUDE_TASKS": "Defines a comma-separated list of test procedure tasks to run. By default, all tasks listed in a test procedure array are run. See https://opensearch.org/docs/latest/benchmark/reference/commands/command-flags/#include-tasks", + "TEST_PROCEDURE": "Defines a test procedure to use. e.g., `append-no-conflicts,significant-text`. Uses default if none provided. See https://opensearch.org/docs/latest/benchmark/reference/commands/command-flags/#test-procedure", + "CAPTURE_NODE_STAT": "Enable opensearch-benchmark node-stats telemetry to capture system level metrics like cpu, jvm etc., see https://opensearch.org/docs/latest/benchmark/reference/telemetry/#node-stats" + }, + "cluster_configuration": { + "size": "Single-Node/Multi-Node", + "data_instance_config": "data-instance-config, e.g., 4vCPU, 32G Mem, 16G Heap" + } + } +} +``` +To add a new test configuration that are suitable to your changes please create a new PR to add desired cluster and benchmark configurations. + +## How to compare my results against baseline? + +Apart from just running benchmarks the user will also be interested in how their change is performing against current OpenSearch distribution with the exactly same cluster and benchmark configurations. +The user can refer to https://s12d.com/basline-dashboards (WIP) to access baseline data for their workload, this data is generated by our nightly benchmark runs on latest build distribution artifacts for 3.0 and 2.x. +In the future, we will add the [compare](https://opensearch.org/docs/latest/benchmark/reference/commands/compare/) feature of opensearch-benchmark to run comparison and publish data on the PR as well. + +## Notes + +- Ensure all required secrets (e.g., `GITHUB_TOKEN`, `UPLOAD_ARCHIVE_ARTIFACT_ROLE`, `ARCHIVE_ARTIFACT_BUCKET_NAME`, `JENKINS_PR_BENCHMARK_GENERIC_WEBHOOK_TOKEN`) are properly set in the repository secrets. +- The `CODEOWNERS` file should list the GitHub usernames of approvers for the benchmark process. + +By following these instructions, repository maintainers can ensure consistent and automated performance benchmarking for all code changes introduced via pull requests. + +