fastapi-alertengine 1.0.3__tar.gz → 1.0.4__tar.gz

{fastapi_alertengine-1.0.3 → fastapi_alertengine-1.0.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fastapi-alertengine
-Version: 1.0.3
+Version: 1.0.4
 Summary: FastAPI AlertEngine: fintech-grade request monitoring (p95 latency + error rate) without Prometheus.
 Author: Tandem Media
 Project-URL: Homepage, https://github.com/Tandem-Media/fastapi-alertengine

fastapi_alertengine-1.0.4/fastapi_alertengine/engine.py

@@ -0,0 +1,114 @@
+# fastapi_alertengine/engine.py
+
+import time
+from typing import Dict, Any
+
+
+STREAM_KEY = "anchorflow:request_metrics"
+
+
+class AlertEngine:
+    """
+    Real-time SLO / latency alert engine.
+
+    Uses rolling window analysis over Redis Stream data.
+    """
+
+    def __init__(self, redis):
+        self.redis = redis
+
+    def _fetch_recent(self, last_n: int = 200):
+        raw = self.redis.xrevrange(STREAM_KEY, count=last_n)
+
+        events = []
+        for _, fields in raw:
+            try:
+                events.append({
+                    "latency_ms": float(fields.get("latency_ms", 0)),
+                    "type": fields.get("type", "api"),
+                    "status_code": int(fields.get("status_code", 0)),
+                    "timestamp": int(fields.get("timestamp", 0)),
+                })
+            except Exception:
+                continue
+
+        return events
+
+    def _p95(self, values):
+        if not values:
+            return 0.0
+
+        values.sort()
+        idx = int(len(values) * 0.95)
+        return values[min(idx, len(values) - 1)]
+
+    def _anomaly_score(self, current, baseline):
+        if baseline == 0:
+            return 0
+
+        return abs(current - baseline) / baseline
+
+    def evaluate(self, window_size: int = 200) -> Dict[str, Any]:
+        events = self._fetch_recent(window_size)
+
+        if not events:
+            return {
+                "status": "ok",
+                "reason": "no_data"
+            }
+
+        all_lat = [e["latency_ms"] for e in events]
+        webhook_lat = [e["latency_ms"] for e in events if e["type"] == "webhook"]
+        api_lat = [e["latency_ms"] for e in events if e["type"] == "api"]
+
+        def p95(values):
+            if not values:
+                return 0.0
+            values.sort()
+            idx = int(len(values) * 0.95)
+            return values[min(idx, len(values) - 1)]
+
+        overall_p95 = p95(all_lat)
+        webhook_p95 = p95(webhook_lat)
+        api_p95 = p95(api_lat)
+
+        # Baseline (simple rolling heuristic)
+        baseline = sum(all_lat) / len(all_lat)
+
+        anomaly = self._anomaly_score(overall_p95, baseline)
+
+        # Alert thresholds
+        status = "ok"
+
+        if overall_p95 > 3000 or anomaly > 2.0:
+            status = "critical"
+        elif overall_p95 > 1000 or anomaly > 1.0:
+            status = "warning"
+
+        # Failure pressure signal
+        error_rate = sum(1 for e in events if e["status_code"] >= 500) / len(events)
+
+        if error_rate > 0.2:
+            status = "critical"
+        elif error_rate > 0.1 and status != "critical":
+            status = "warning"
+
+        return {
+            "status": status,
+            "metrics": {
+                "overall_p95_ms": overall_p95,
+                "webhook_p95_ms": webhook_p95,
+                "api_p95_ms": api_p95,
+                "error_rate": error_rate,
+                "anomaly_score": anomaly,
+                "sample_size": len(events),
+            },
+            "thresholds": {
+                "p95_warning": 1000,
+                "p95_critical": 3000,
+                "anomaly_warning": 1.0,
+                "anomaly_critical": 2.0,
+                "error_rate_critical": 0.2,
+            },
+            "timestamp": int(time.time()),
+        }

{fastapi_alertengine-1.0.3 → fastapi_alertengine-1.0.4}/fastapi_alertengine.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fastapi-alertengine
-Version: 1.0.3
+Version: 1.0.4
 Summary: FastAPI AlertEngine: fintech-grade request monitoring (p95 latency + error rate) without Prometheus.
 Author: Tandem Media
 Project-URL: Homepage, https://github.com/Tandem-Media/fastapi-alertengine

{fastapi_alertengine-1.0.3 → fastapi_alertengine-1.0.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "fastapi-alertengine"
-version = "1.0.3"
+version = "1.0.4"
 description = "FastAPI AlertEngine: fintech-grade request monitoring (p95 latency + error rate) without Prometheus."
 readme = "README.md"
 requires-python = ">=3.10"

{fastapi_alertengine-1.0.3 → fastapi_alertengine-1.0.4}/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = fastapi-alertengine
-version = 1.0.3
+version = "1.0.4"
 
 [options]
 packages = find: