init stacks

14_infra-assistant/api/correlation/analyzer.py

@@ -0,0 +1,31 @@
+def analyze(metrics: dict):
+    alerts = []
+
+    if metrics.get("cpu"):
+        cpu = float(metrics["cpu"][0]["value"][1])
+        if cpu > 85:
+            alerts.append({
+                "title": "CPU élevée",
+                "cause": f"{cpu:.1f} %",
+                "severity": "warning",
+            })
+
+    if metrics.get("memory"):
+        mem = float(metrics["memory"][0]["value"][1])
+        if mem > 90:
+            alerts.append({
+                "title": "Mémoire saturée",
+                "cause": f"{mem:.1f} %",
+                "severity": "critical",
+            })
+
+    if metrics.get("containers_down"):
+        down = int(float(metrics["containers_down"][0]["value"][1]))
+        if down > 0:
+            alerts.append({
+                "title": "Containers arrêtés",
+                "cause": f"{down} détecté(s)",
+                "severity": "critical",
+            })
+
+    return alerts

14_infra-assistant/api/correlation/collectors.py

@@ -0,0 +1,14 @@
+from api.connectors.prometheus import (
+    get_cpu_usage,
+    get_memory_usage,
+    get_disk_usage,
+    get_container_down,
+)
+
+def collect_metrics():
+    return {
+        "cpu": get_cpu_usage(),
+        "memory": get_memory_usage(),
+        "disk": get_disk_usage(),
+        "containers_down": get_container_down(),
+    }

14_infra-assistant/api/correlation/engine.py

@@ -0,0 +1,44 @@
+import yaml
+
+with open("correlation/rules.yaml") as f:
+    RULES = yaml.safe_load(f)
+
+def correlate(alert):
+    service = alert["service"]
+    rules = RULES["services"].get(service, {})
+    return {
+        "service": service,
+        "severity": alert["severity"],
+        "metrics": alert["metrics"],
+        "depends_on": rules.get("depends_on", []),
+        "impact": rules.get("impact", [])
+    }
+
+def evaluate(rules: list, metrics: dict):
+    incidents = []
+
+    for rule in rules:
+        matched = True
+
+        for metric, condition in rule["when"].items():
+            value = metrics.get(metric)
+
+            if value is None:
+                matched = False
+                break
+
+            if "gt" in condition and value <= condition["gt"]:
+                matched = False
+
+            if "lt" in condition and value >= condition["lt"]:
+                matched = False
+
+        if matched:
+            incidents.append({
+                "id": rule["id"],
+                "severity": rule["severity"],
+                "impact": rule["impact"],
+                "metrics": metrics,
+            })
+
+    return incidents

14_infra-assistant/api/correlation/rules.py

@@ -0,0 +1,14 @@
+import yaml
+from pathlib import Path
+
+RULES_FILE = Path(__file__).parent / "rules.yaml"
+
+
+def load_rules():
+    if not RULES_FILE.exists():
+        return []
+
+    with RULES_FILE.open("r") as f:
+        data = yaml.safe_load(f) or {}
+
+    return data.get("rules", [])

14_infra-assistant/api/correlation/rules.yaml

@@ -0,0 +1,34 @@
+services:
+  jellyfin:
+    depends_on:
+      - tdarr
+      - cpu
+    impact:
+      - media
+      - nextcloud
+
+  nextcloud:
+    depends_on:
+      - database
+      - redis
+
+rules:
+  - id: cpu_ram_pressure
+    description: CPU élevé + mémoire sous pression
+    when:
+      cpu:
+        gt: 80
+        for: 5m
+      memory_available:
+        lt: 15
+    impact: "Risque fort de ralentissement général"
+    severity: high
+
+  - id: io_wait_saturation
+    description: IO wait élevé
+    when:
+      io_wait:
+        gt: 20
+        for: 3m
+    impact: "Disque saturé, latences applicatives"
+    severity: high