snapshot: add namespace usage stats

ariadne/services/cluster_state.py

@@ -445,6 +445,21 @@ def _vm_vector(expr: str) -> list[dict[str, Any]]:
     return output
 
 
+def _filter_namespace_vector(entries: list[dict[str, Any]]) -> list[dict[str, Any]]:
+    output: list[dict[str, Any]] = []
+    for item in entries:
+        if not isinstance(item, dict):
+            continue
+        metric = item.get("metric") if isinstance(item.get("metric"), dict) else {}
+        namespace = metric.get("namespace")
+        if not isinstance(namespace, str) or not namespace:
+            continue
+        if namespace in _SYSTEM_NAMESPACES:
+            continue
+        output.append(item)
+    return output
+
+
 def _vm_topk(expr: str, label_key: str) -> dict[str, Any] | None:
     result = _vm_vector(expr)
     if not result:
@@ -538,6 +553,24 @@ def _node_usage(errors: list[str]) -> dict[str, Any]:
     return usage
 
 
+def _usage_stats(series: list[dict[str, Any]]) -> dict[str, float]:
+    values: list[float] = []
+    for entry in series:
+        if not isinstance(entry, dict):
+            continue
+        try:
+            values.append(float(entry.get("value")))
+        except (TypeError, ValueError):
+            continue
+    if not values:
+        return {}
+    return {
+        "min": min(values),
+        "max": max(values),
+        "avg": sum(values) / len(values),
+    }
+
+
 def _summarize_metrics(errors: list[str]) -> dict[str, Any]:
     metrics: dict[str, Any] = {}
     try:
@@ -557,12 +590,33 @@ def _summarize_metrics(errors: list[str]) -> dict[str, Any]:
     metrics["postgres_connections"] = _postgres_connections(errors)
     metrics["hottest_nodes"] = _hottest_nodes(errors)
     metrics["node_usage"] = _node_usage(errors)
+    metrics["node_usage_stats"] = {
+        "cpu": _usage_stats(metrics.get("node_usage", {}).get("cpu", [])),
+        "ram": _usage_stats(metrics.get("node_usage", {}).get("ram", [])),
+        "net": _usage_stats(metrics.get("node_usage", {}).get("net", [])),
+        "io": _usage_stats(metrics.get("node_usage", {}).get("io", [])),
+    }
+    try:
+        metrics["namespace_cpu_top"] = _filter_namespace_vector(
+            _vm_vector(
+                f'topk(5, sum by (namespace) (rate(container_cpu_usage_seconds_total{{namespace!=""}}[{_RATE_WINDOW}])))'
+            )
+        )
+        metrics["namespace_mem_top"] = _filter_namespace_vector(
+            _vm_vector(
+                "topk(5, sum by (namespace) (container_memory_working_set_bytes{namespace!=\"\"}))"
+            )
+        )
+    except Exception as exc:
+        errors.append(f"namespace_usage: {exc}")
     metrics["units"] = {
         "cpu": "percent",
         "ram": "percent",
         "net": "bytes_per_sec",
         "io": "bytes_per_sec",
         "restarts": "count",
+        "namespace_cpu": "cores",
+        "namespace_mem": "bytes",
     }
     metrics["windows"] = {
         "rates": _RATE_WINDOW,

tests/test_cluster_state.py

@@ -82,6 +82,9 @@ def test_collect_cluster_state(monkeypatch) -> None:
     assert snapshot["workloads"]
     assert snapshot["namespace_pods"]
     assert snapshot["namespace_pods"][0]["namespace"] == "media"
+    assert "node_usage_stats" in snapshot["metrics"]
+    assert snapshot["metrics"]["namespace_cpu_top"] == []
+    assert snapshot["metrics"]["namespace_mem_top"] == []
     assert summary.nodes_total == 2
     assert summary.nodes_ready == 1
     assert summary.pods_running == 5.0