monitoring: bound atlas availability query

scripts/dashboards_render_atlas.py

@@ -312,7 +312,7 @@ STUCK_TERMINATING_EXPR = (
     ')) '
     "or on() vector(0)"
 )
-UPTIME_WINDOW = "365d"
+UPTIME_WINDOW = "30d"
 # Keep the subquery step coarse so we don't request an excessive number of points.
 UPTIME_STEP = "1h"
 TRAEFIK_READY_EXPR = (
@@ -1433,13 +1433,15 @@ def build_overview():
         },
         {
             "id": 27,
-            "title": "Atlas Availability",
+            "title": "Atlas Availability (30d)",
             "expr": UPTIME_PERCENT_EXPR,
             "kind": "stat",
             "thresholds": UPTIME_PERCENT_THRESHOLDS,
             "unit": "percentunit",
             "decimals": 4,
             "text_mode": "value",
+            "instant": True,
+            "description": "Rolling 30-day availability from control-plane readiness and Traefik readiness. Kept to 30d so missing pre-metric history is not counted as downtime.",
         },
         {
             "id": 4,
@@ -1509,14 +1511,16 @@ def build_overview():
                     item["expr"],
                     grid,
                     thresholds=item.get("thresholds"),
-            legend=None,
+                    legend=None,
-            links=item.get("links"),
+                    links=item.get("links"),
-            text_mode=item.get("text_mode", "value"),
+                    text_mode=item.get("text_mode", "value"),
-            value_suffix=item.get("value_suffix"),
+                    value_suffix=item.get("value_suffix"),
-            unit=item.get("unit", "none"),
+                    unit=item.get("unit", "none"),
-            decimals=item.get("decimals"),
+                    decimals=item.get("decimals"),
-        )
+                    instant=item.get("instant", False),
-    )
+                    description=item.get("description"),
+                )
+            )
         else:
             panels.append(
                 gauge_panel(

scripts/tests/test_dashboards_render_atlas.py

@@ -50,6 +50,18 @@ def test_node_filter_and_expr_helpers():
     assert "node_memory_MemAvailable_bytes" in mem_expr
 
 
+def test_overview_availability_panel_is_recent_and_instant():
+    mod = load_module()
+    dashboard = mod.build_overview()
+    panel = next(panel for panel in flatten_panels(dashboard["panels"]) if panel["id"] == 27)
+
+    assert panel["title"] == "Atlas Availability (30d)"
+    assert "[30d:1h]" in panel["targets"][0]["expr"]
+    assert "365d" not in panel["targets"][0]["expr"]
+    assert panel["targets"][0]["instant"] is True
+    assert "pre-metric history" in panel["description"]
+
+
 def test_render_configmap_writes(tmp_path):
     mod = load_module()
     mod.DASHBOARD_DIR = tmp_path / "dash"

services/monitoring/dashboards/atlas-overview.json

@@ -213,7 +213,7 @@
     {
       "id": 27,
       "type": "stat",
-      "title": "Atlas Availability",
+      "title": "Atlas Availability (30d)",
       "datasource": {
         "type": "prometheus",
         "uid": "atlas-vm"
@@ -226,8 +226,9 @@
       },
       "targets": [
         {
-          "expr": "avg_over_time((min(((sum(kube_node_status_condition{condition=\"Ready\",status=\"true\",node=~\"titan-0a|titan-0b|titan-0c\"}) / 3)), ((sum(kube_deployment_status_replicas_available{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}) / clamp_min(sum(kube_deployment_spec_replicas{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}), 1)))))[365d:1h])",
+          "expr": "avg_over_time((min(((sum(kube_node_status_condition{condition=\"Ready\",status=\"true\",node=~\"titan-0a|titan-0b|titan-0c\"}) / 3)), ((sum(kube_deployment_status_replicas_available{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}) / clamp_min(sum(kube_deployment_spec_replicas{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}), 1)))))[30d:1h])",
-          "refId": "A"
+          "refId": "A",
+          "instant": true
         }
       ],
       "fieldConfig": {
@@ -281,7 +282,8 @@
           "values": false
         },
         "textMode": "value"
-      }
+      },
+      "description": "Rolling 30-day availability from control-plane readiness and Traefik readiness. Kept to 30d so missing pre-metric history is not counted as downtime."
     },
     {
       "id": 4,

services/monitoring/grafana-dashboard-overview.yaml

@@ -222,7 +222,7 @@ data:
         {
           "id": 27,
           "type": "stat",
-          "title": "Atlas Availability",
+          "title": "Atlas Availability (30d)",
           "datasource": {
             "type": "prometheus",
             "uid": "atlas-vm"
@@ -235,8 +235,9 @@ data:
           },
           "targets": [
             {
-              "expr": "avg_over_time((min(((sum(kube_node_status_condition{condition=\"Ready\",status=\"true\",node=~\"titan-0a|titan-0b|titan-0c\"}) / 3)), ((sum(kube_deployment_status_replicas_available{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}) / clamp_min(sum(kube_deployment_spec_replicas{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}), 1)))))[365d:1h])",
+              "expr": "avg_over_time((min(((sum(kube_node_status_condition{condition=\"Ready\",status=\"true\",node=~\"titan-0a|titan-0b|titan-0c\"}) / 3)), ((sum(kube_deployment_status_replicas_available{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}) / clamp_min(sum(kube_deployment_spec_replicas{namespace=~\"traefik|kube-system\",deployment=\"traefik\"}), 1)))))[30d:1h])",
-              "refId": "A"
+              "refId": "A",
+              "instant": true
             }
           ],
           "fieldConfig": {
@@ -290,7 +291,8 @@ data:
               "values": false
             },
             "textMode": "value"
-          }
+          },
+          "description": "Rolling 30-day availability from control-plane readiness and Traefik readiness. Kept to 30d so missing pre-metric history is not counted as downtime."
         },
         {
           "id": 4,