linux memory subsystem update

app.py

@@ -16,9 +16,29 @@
 import shutil
 from datetime import datetime
 from threading import Lock
-from flask import Flask, jsonify, render_template, send_from_directory, request, redirect
+from flask import Flask, jsonify, render_template, send_from_directory, request, redirect, g, Response
 import psutil
 
+# Gunicorn -w N: set PROMETHEUS_MULTIPROC_DIR before workers import this module (e.g. in gunicorn.conf.py).
+if os.environ.get("PROMETHEUS_MULTIPROC_DIR", "").strip():
+    _prom_mpdir = os.environ["PROMETHEUS_MULTIPROC_DIR"].strip()
+    os.makedirs(_prom_mpdir, exist_ok=True)
+
+try:
+    from prometheus_client import (
+        CONTENT_TYPE_LATEST,
+        CollectorRegistry,
+        Counter,
+        Histogram,
+        generate_latest,
+        multiprocess,
+    )
+
+    _PROMETHEUS_AVAILABLE = True
+except ImportError:
+    _PROMETHEUS_AVAILABLE = False
+    CONTENT_TYPE_LATEST = "text/plain; version=0.0.4; charset=utf-8"
+
 # Try to import OpenAI (optional)
 try:
     import openai
@@ -27,6 +47,60 @@
     OPENAI_AVAILABLE = False
 
 app = Flask(__name__)
+
+if _PROMETHEUS_AVAILABLE:
+    REQUEST_COUNT = Counter(
+        "http_requests_total",
+        "Total HTTP requests",
+        ["method", "endpoint", "status"],
+    )
+    REQUEST_LATENCY = Histogram(
+        "http_request_duration_seconds",
+        "HTTP request latency in seconds",
+        buckets=(0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1.0, 2.5, 5.0, 10.0, float("inf")),
+    )
+
+    @app.before_request
+    def _prometheus_before_request():
+        g._prom_start = time.perf_counter()
+
+    @app.after_request
+    def _prometheus_after_request(response):
+        start = getattr(g, "_prom_start", None)
+        if start is not None:
+            REQUEST_LATENCY.observe(time.perf_counter() - start)
+        ep = request.endpoint
+        rule = request.url_rule.rule if request.url_rule else None
+        endpoint_label = ep or rule or "unmatched"
+        try:
+            REQUEST_COUNT.labels(
+                method=request.method,
+                endpoint=endpoint_label,
+                status=str(response.status_code),
+            ).inc()
+        except Exception:
+            pass
+        return response
+
+    @app.route("/metrics")
+    def prometheus_metrics():
+        if os.environ.get("PROMETHEUS_MULTIPROC_DIR", "").strip():
+            registry = CollectorRegistry()
+            multiprocess.MultiProcessCollector(registry)
+            data = generate_latest(registry)
+        else:
+            data = generate_latest()
+        return Response(data, mimetype=CONTENT_TYPE_LATEST)
+
+else:
+
+    @app.route("/metrics")
+    def prometheus_metrics_disabled():
+        return jsonify(
+            {"error": "prometheus_client not installed; pip install prometheus-client"}
+        ), 503
+
+
 TRACEROUTE_CACHE = {}
 TRACEROUTE_CACHE_TTL_SECONDS = 60
 NETWORK_STACK_PREV = {

templates/linux-memory-subsystem.html

@@ -87,7 +87,7 @@ <h2>Linux kernel memory management</h2>
 
     <script src="/static/js/safe-utils.js?v=2"></script>
     <script src="/static/js/frontend-logger.js?v=2"></script>
-    <script src="/static/js/memory-belt.js?v=1"></script>
+    <script src="/static/js/memory-belt.js?v=2"></script>
     <script>
         (function startMemoryPage() {
             if (typeof window.MemorySubsystemVisualization !== 'function') {