Enhance observability: wire up metrics for reload count, access log drops, and tracing spans

Author: fabian4

.github/workflows/pipeline.yaml

@@ -293,7 +293,7 @@ main() {
   pod_label=$(get_proxy_pod_label)
   proxy_port=$(get_proxy_port)
 
-  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}"
+  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}${BENCH_CASE_SUFFIX:+__${BENCH_CASE_SUFFIX}}"
   ensure_dir "$output_dir"
 
   # Setup port-forward to access test backend (for load generation only)

Cargo.lock

@@ -36,7 +36,7 @@ main() {
   container_name=$(get_proxy_container_name)
   proxy_port=$(get_proxy_port)
 
-  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}"
+  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}${BENCH_CASE_SUFFIX:+__${BENCH_CASE_SUFFIX}}"
   ensure_dir "$output_dir"
 
   # Setup port-forward to access test backend

bench/cases/system/config_reload_convergence.sh

@@ -151,7 +151,7 @@ main() {
   pod_label=$(get_proxy_pod_label)
   proxy_port=$(get_proxy_port)
 
-  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}"
+  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}${BENCH_CASE_SUFFIX:+__${BENCH_CASE_SUFFIX}}"
   ensure_dir "$output_dir"
 
   # Setup port-forward to access test backend

bench/cases/system/multi_hour_soak.sh

@@ -131,7 +131,7 @@ main() {
   container_name=$(get_proxy_container_name)
   proxy_port=$(get_proxy_port)
 
-  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}"
+  local output_dir="${BENCH_OUTPUT_DIR}/${BENCH_MODE}/${BENCH_PROXY}/${CASE_NAME}${BENCH_CASE_SUFFIX:+__${BENCH_CASE_SUFFIX}}"
   ensure_dir "$output_dir"
 
   # Setup port-forward to access test backend

bench/cases/system/rollback_performance.sh

@@ -8,6 +8,7 @@ telemetry:
   pingora: "error"
   service_name: "pavis-benchmark"
   access_log: false
+  metrics: "0.0.0.0:9090"
 
 upstreams:
   - name: "backend"

bench/cases/system/stress_recovery.sh

@@ -8,6 +8,7 @@ telemetry:
   pingora: "error"
   service_name: "pavis-benchmark"
   access_log: false
+  metrics: "0.0.0.0:9090"
 
 upstreams:
   - name: "backend"

bench/config/standalone/pavis.yaml

@@ -65,6 +65,7 @@ data:
       pingora: "error"
       service_name: "pavis-benchmark"
       access_log: false
+      metrics: "0.0.0.0:9090"
 
     upstreams:
       - name: "backend"

bench/config/system/pavis/pavis.yaml

@@ -73,6 +73,9 @@ spec:
         - name: proxy
           containerPort: 8080
           protocol: TCP
+        - name: metrics
+          containerPort: 9090
+          protocol: TCP
         env:
         - name: PAVIS_RELAY_URL
           value: "http://pavis-relay:8090"