Add Process Resources section to Grafana dashboard

Add row with panels for memory usage (RSS/virtual), CPU usage,
file descriptors and threads monitoring.

Also clarify rejection scenario (503) and how to overcome
effective rate degradation in high load testing.

Author: testillano

README.md

@@ -639,6 +639,22 @@ Also, reports are generated as markdown files under the profile's `reports/` sub
 * Data storage is disabled in the script by default to prevent memory from growing and improve server response times (remember that storage shall be kept when provisions require data persistence).
 * In general, even with high traffic rates, you could get sneaky snapshots just enabling and then quickly disabling data storage, for example using [function helpers](#Helper-functions): `server_data_configuration --keep-all && server_data_configuration --discard-all`
 
+#### Handling 503 rejections in client mode
+
+When using h2agent as a traffic client against a server under congestion, 503 responses indicate the server cannot sustain the requested rate.
+
+**Breaking the chain on rejection** (detecting the problem):
+
+* **`expectedResponseStatusCode`**: set the expected value (e.g., `201`). If the actual response differs, the chain breaks immediately. Simple and deterministic.
+* **`onResponseTransform` with filters**: read `response.statusCode` as source, apply conditional logic via filters. On mismatch, use `onFilterFail` to set an error `outState`, write diagnostic vaults, or trigger alternative flows. More flexible than the binary kill-switch above.
+* **Omit `expectedResponseStatusCode`**: the chain continues regardless of status code. Useful when you want to collect all events for post-hoc analysis without interrupting the flow.
+
+**Managing load under congestion** (mitigating the problem):
+
+* **External rate controller**: monitor the `h2agent_traffic_client_observed_responses_received_counter{status_code="503"}` metric and adjust the trigger rate dynamically (e.g., a PID loop that reduces CPS when rejection rate exceeds a threshold, or compensates effective load by increasing rate to offset rejections).
+* **Tiered rate ramp**: start at low rate and increase stepwise, measuring sustainable throughput at each plateau. Stop when 503 rate exceeds acceptable threshold -- this measures Engineering Capacity (EC).
+* **Long ramp-up**: increase rate gradually over time to let the server warm up and stabilize, reducing transient rejections during rate transitions.
+
 
 
 So you may start the process, again, natively or using docker:

build-native.sh

@@ -17,10 +17,10 @@ boost_ver=1.84.0 # safer to have this version (https://github.com/nghttp2/nghttp
 ert_nghttp2_ver=v1.2.9 # to download nghttp2 patches (this must be aligned with previous: nghttp2 & nghttp2-asio & boost)
 ert_logger_ver=v1.1.1
 ert_queuedispatcher_ver=v1.0.4
-jupp0r_prometheuscpp_ver=v0.13.0
-civetweb_civetweb_ver=v1.14
-ert_metrics_ver=v1.1.1
-ert_http2comm_ver=v2.2.15
+jupp0r_prometheuscpp_ver=v1.3.0
+civetweb_civetweb_ver=v1.16
+ert_metrics_ver=v1.2.0
+ert_http2comm_ver=v2.3.0
 nlohmann_json_ver=$(grep ^nlohmann_json_ver__dflt= ${project_root_dir}/build.sh | cut -d= -f2)
 pboettch_jsonschemavalidator_ver=$(grep ^pboettch_jsonschemavalidator_ver__dflt= ${project_root_dir}/build.sh | cut -d= -f2)
 google_test_ver=$(grep ^google_test_ver__dflt= ${project_root_dir}/build.sh | cut -d= -f2)

src/main.cpp

@@ -1083,6 +1083,7 @@ ChatGPT:        https://github.com/testillano/h2agent/blob/master/README.md#ques
             std::cerr << currentDateTime() << ": Initialization error in prometheus interface (" << bind_address_port_prometheus_exposer << "). Exiting ..." << '\n';
             myExit(EXIT_FAILURE);
         }
+        myMetrics->enableProcessMetrics(application_name);
     }
 
     // FileManager/SafeFile metrics