Claude/monitor worker improvements x vl95

workers/arax/worker.py

@@ -31,8 +31,10 @@ async def arax(task, logger: logging.Logger):
         logger.info(f"Get the message from db {message}")
 
         headers = {"Content-Type": "application/json"}
-        with httpx.Client(timeout=270) as client:
-            response = client.post(settings.arax_url, json=message, headers=headers)
+        async with httpx.AsyncClient(timeout=270) as client:
+            response = await client.post(
+                settings.arax_url, json=message, headers=headers
+            )
 
         logger.info(f"Status Code from ARAX response: {response.status_code}")
         result = response.json()

workers/monitor/alerts.py

@@ -306,8 +306,14 @@ async def _dispatch_slack(event: Dict[str, Any]) -> None:
         "warning": ":warning:",
         "critical": ":rotating_light:",
     }.get(event.get("severity", "warning"), ":warning:")
+    # Environment context lets one Slack channel receive alerts from multiple
+    # deployments (dev / staging / production) without ambiguity about which
+    # one fired.
+    server_url = settings.server_url or "unknown"
+    maturity = settings.server_maturity or "unknown"
     text = (
         f"{emoji} *Shepherd alert* `{event['rule']}` ({event['severity']})\n"
+        f"*Environment:* {maturity}  |  *URL:* <{server_url}|{server_url}>\n"
         f"{event['message']}"
     )
     try:

workers/monitor/static/app.js

@@ -22,6 +22,10 @@
   let xlenChart, araChart;
   let xlenHistory = {}; // stream -> [{x, y}, ...]
   const HISTORY_POINTS = 60; // rolling window in the chart
+  // Tracks chart legend items the user clicked to hide. Datasets are rebuilt
+  // on every snapshot tick, which wipes Chart.js's internal visibility state,
+  // so we mirror it here (keyed by series label) and re-apply on each rebuild.
+  const xlenHiddenSeries = new Set();
 
   function fmt(n) {
     if (n === null || n === undefined) return "-";
@@ -56,7 +60,18 @@
           animation: false,
           parsing: false,
           plugins: {
-            legend: { labels: { color: "#8b949e", boxWidth: 10 } },
+            legend: {
+              labels: { color: "#8b949e", boxWidth: 10 },
+              onClick: (e, legendItem, legend) => {
+                Chart.defaults.plugins.legend.onClick(e, legendItem, legend);
+                const label = legendItem.text;
+                if (legend.chart.isDatasetVisible(legendItem.datasetIndex)) {
+                  xlenHiddenSeries.delete(label);
+                } else {
+                  xlenHiddenSeries.add(label);
+                }
+              },
+            },
             title: { display: true, text: "Queue depth (XLEN)", color: "#e6edf3" },
           },
           scales: {
@@ -231,6 +246,7 @@
       borderWidth: 1.5,
       pointRadius: 0,
       tension: 0.2,
+      hidden: xlenHiddenSeries.has(n),
     }));
     xlenChart.update("none");
   }

workers/monitor/storage.py

@@ -395,22 +395,28 @@ async def query_summary(since: float, until: float) -> Dict[str, Any]:
             row = await cur.fetchone()
             summary["queries_started"] = int(row[0] or 0)
 
-            # Event counts.
+            # Event counts. Crashes are not their own row type -- the poller
+            # records them as scale_down events with ``payload.kind = "crashed"``
+            # (a clean scale-down uses kind="scaled_down"), so we have to dig
+            # into the JSON payload to separate the two.
             cur = await conn.execute(
-                "SELECT type, COUNT(*) FROM monitor_events "
+                "SELECT type, COUNT(*) FILTER (WHERE payload->>'kind' = 'crashed'), "
+                "COUNT(*) FROM monitor_events "
                 "WHERE ts >= to_timestamp(%s) AND ts < to_timestamp(%s) "
                 "GROUP BY type",
                 (since, until),
             )
-            for etype, count in await cur.fetchall():
-                if etype == "crash":
-                    summary["crashes"] = int(count)
-                elif etype in ("scale_up", "scale_down"):
+            for etype, crashed_count, total_count in await cur.fetchall():
+                if etype in ("scale_up", "scale_down"):
                     summary["scale_events"] = summary.get("scale_events", 0) + int(
-                        count
+                        total_count
                     )
+                    if etype == "scale_down":
+                        summary["crashes"] = summary.get("crashes", 0) + int(
+                            crashed_count or 0
+                        )
                 elif etype == "alert":
-                    summary["alert_count"] = int(count)
+                    summary["alert_count"] = int(total_count)
 
             # Peak backlog per stream.
             cur = await conn.execute(