ci(benchmark): detect machine with IMDSv2 [skip ci] (#10876)

Author: everpcpc

.github/actions/benchmark_cloud/action.yml

@@ -65,7 +65,6 @@ runs:
       working-directory: benchmark/clickbench
       shell: bash
       run: |
-        yq -i ".date = \"$(date -u +%Y-%m-%d)\"" result.json
         case ${{ inputs.source }} in
           pr)
             yq -i '.system = "Databend(PR#${{ inputs.source_id }})"' result.json
@@ -77,22 +76,7 @@ runs:
             echo "Unspported benchmark source: ${{ inputs.source }}"
             exit 1
         esac
-
-        case ${{ inputs.size }} in
-          Medium)
-            yq -i '.cluster_size = "16"' result.json
-            yq -i '.machine = "16×Medium"' result.json
-            ;;
-          Large)
-            yq -i '.cluster_size = "64"' result.json
-            yq -i '.machine = "64×Large"' result.json
-            ;;
-          *)
-            echo "Unspported benchmark size: ${{ inputs.size }}"
-            exit 1
-        esac
-        yq -i '.tags = ["s3"]' result.json
-        yq -i '.comment = "commit:${{ inputs.sha }}"' result.json
+        yq -i '.comment = "commit: ${{ inputs.sha }}"' result.json
         mv result.json result-${{ inputs.dataset }}-cloud-${{ inputs.size }}.json
 
     - name: Upload artifact

.github/actions/benchmark_local/action.yml

@@ -50,12 +50,7 @@ runs:
             echo "Unspported benchmark source: ${{ inputs.source }}"
             exit 1
         esac
-        yq -i ".date = \"$(date -u +%Y-%m-%d)\"" result.json
-        yq -i '.machine = "c5.4xlarge"' result.json
-        yq -i '.cluster_size = 1' result.json
-        yq -i '.comment = "commit:${{ inputs.sha }}"' result.json
-        yq -i '.tags = ["gp3"]' result.json
-        yq -i '.extra.${{ inputs.source }} = "${{ inputs.source_id }}"' result.json
+        yq -i '.comment = "commit: ${{ inputs.sha }}"' result.json
         mv result.json result-${{ inputs.dataset }}-local.json
 
     - name: Upload artifact

benchmark/clickbench/benchmark_cloud.sh

@@ -29,6 +29,25 @@ python3 --version
 yq --version
 bendsql version
 
+echo "Preparing benchmark metadata..."
+echo '{}' >result.json
+yq -i ".date = \"$(date -u +%Y-%m-%d)\"" result.json
+yq -i '.tags = ["s3"]' result.json
+case ${BENCHMARK_SIZE} in
+Medium)
+    yq -i '.cluster_size = "16"' result.json
+    yq -i '.machine = "16×Medium"' result.json
+    ;;
+Large)
+    yq -i '.cluster_size = "64"' result.json
+    yq -i '.machine = "64×Large"' result.json
+    ;;
+*)
+    echo "Unspported benchmark size: ${BENCHMARK_SIZE}"
+    exit 1
+    ;;
+esac
+
 echo "#######################################################"
 echo "Running benchmark for Databend Cloud with S3 storage..."
 
@@ -44,10 +63,8 @@ bendsql cloud warehouse ls
 bendsql cloud warehouse create "${CLOUD_WAREHOUSE}" --size "${BENCHMARK_SIZE}" --tag "${BENCHMARK_IMAGE_TAG}"
 bendsql cloud warehouse ls
 bendsql cloud warehouse resume "${CLOUD_WAREHOUSE}" --wait
-
 bendsql cloud warehouse use "${CLOUD_WAREHOUSE}"
 
-echo '{}' >result.json
 echo "Running queries..."
 
 function run_query() {

benchmark/clickbench/benchmark_local.sh

@@ -47,6 +47,11 @@ echo "CREATE DATABASE ${BENCHMARK_DATASET};" | bendsql query
 echo "Creating table for benchmark with native storage format..."
 bendsql query <"${BENCHMARK_DATASET}/create_local.sql"
 
+# Detect instance type with AWS metadata
+token=$(curl -X PUT "http://169.254.169.254/latest/api/token" -H "X-aws-ec2-metadata-token-ttl-seconds: 60")
+instance_type=$(curl -H "X-aws-ec2-metadata-token: $token" http://169.254.169.254/latest/meta-data/instance-type)
+echo "Instance type: ${instance_type}"
+
 echo "Loading data..."
 load_start=$(date +%s)
 bendsql query <"${BENCHMARK_DATASET}/load.sql"
@@ -57,7 +62,11 @@ echo "Data loaded in ${load_time}s."
 data_size=$(echo "select sum(data_compressed_size) from system.tables where database = '${BENCHMARK_DATASET}';" | bendsql query -f unaligned -t)
 
 echo '{}' >result.json
+yq -i ".date = \"$(date -u +%Y-%m-%d)\"" result.json
 yq -i ".load_time = ${load_time} | .data_size = ${data_size} | .result = []" result.json
+yq -i ".machine = \"${instance_type}\"" result.json
+yq -i '.cluster_size = 1' result.json
+yq -i '.tags = ["gp3"]' result.json
 
 echo "Running queries..."
 

benchmark/clickbench/index.jinja

@@ -304,9 +304,9 @@
     <div class="header stick-left">
         <span class="nowrap themes"><span id="toggle-dark">🌚</span><span id="toggle-light">🌞</span></span>
         <h1>{{ title }}</h1>
-        <a href="https://github.com/ClickHouse/ClickBench/">Other System</a> |
-        <a href="https://repo.databend.rs/benchmark/clickbench/release/tpch.html">Release Versions Benchmark</a> |
-        <a href="{{ url }}">Benchmark Source Detail</a>
+        <a href="https://github.com/ClickHouse/ClickBench/">Clickbench</a> |
+        <a href="https://repo.databend.rs/benchmark/clickbench/release/{{ dataset }}.html">Release Versions</a> |
+        <a href="{{ url }}">Changeset Detail</a>
     </div>
 
     <table class="selectors-container stick-left">

benchmark/clickbench/update_results.py

@@ -30,7 +30,13 @@ def update_results(dataset, title, url):
     templateEnv = Environment(loader=templateLoader)
     template = templateEnv.get_template(TEMPLATE_FILE)
     logger.info("rendering result with args: %s ...", args)
-    outputText = template.render(title=title, url=url, queries=queries, results=results)
+    outputText = template.render(
+        dataset=dataset,
+        title=title,
+        url=url,
+        queries=queries,
+        results=results,
+    )
     with open(f"results/{dataset}.html", "w") as f:
         f.write(outputText)