Initial commit — basic SLO eval working

README.md

@@ -0,0 +1,23 @@
+# slo-tracker
+
+Minimal CLI for tracking SLO burn rates from Prometheus.
+Outputs a quick summary to stdout — useful for morning standups and incident reviews.
+
+Built for my own use. Tested on the Nexus prod setup.
+
+## Usage
+
+    python slo_tracker.py --config slos.yml --window 7d
+
+## Config format
+
+    slos:
+      - name: "API availability"
+        query: 'sum(rate(http_requests_total{code!~"5.."}[5m])) / sum(rate(http_requests_total[5m]))'
+        target: 0.999
+        window: 30d
+
+## Requirements
+
+    pip install prometheus-api-client pyyaml rich
+

requirements.txt

@@ -0,0 +1,3 @@
+prometheus-api-client==0.5.4
+pyyaml==6.0.1
+rich==13.7.1

slo_tracker.py

@@ -0,0 +1,64 @@
+#!/usr/bin/env python3
+"""Minimal SLO burn rate tracker for Prometheus."""
+
+import argparse
+import sys
+from datetime import datetime, timedelta
+
+import yaml
+from prometheus_api_client import PrometheusConnect
+from rich.console import Console
+from rich.table import Table
+
+console = Console()
+
+
+def load_config(path: str) -> dict:
+    with open(path) as f:
+        return yaml.safe_load(f)
+
+
+def evaluate_slo(prom: PrometheusConnect, slo: dict, window: str) -> dict:
+    result = prom.custom_query(slo["query"])
+    if not result:
+        return {"name": slo["name"], "value": None, "target": slo["target"], "ok": False}
+    value = float(result[0]["value"][1])
+    return {
+        "name": slo["name"],
+        "value": value,
+        "target": slo["target"],
+        "ok": value >= slo["target"],
+    }
+
+
+def main():
+    parser = argparse.ArgumentParser(description="SLO burn rate tracker")
+    parser.add_argument("--config", required=True)
+    parser.add_argument("--prometheus", default="http://localhost:9090")
+    parser.add_argument("--window", default="7d")
+    args = parser.parse_args()
+
+    cfg = load_config(args.config)
+    prom = PrometheusConnect(url=args.prometheus, disable_ssl=True)
+
+    table = Table(title=f"SLO Status ({args.window} window)")
+    table.add_column("SLO", style="cyan")
+    table.add_column("Target", justify="right")
+    table.add_column("Current", justify="right")
+    table.add_column("Status", justify="center")
+
+    all_ok = True
+    for slo in cfg.get("slos", []):
+        result = evaluate_slo(prom, slo, args.window)
+        status = "[green]OK[/green]" if result["ok"] else "[red]BURNING[/red]"
+        value_str = f"{result['value']:.4f}" if result["value"] is not None else "n/a"
+        table.add_row(result["name"], str(result["target"]), value_str, status)
+        if not result["ok"]:
+            all_ok = False
+
+    console.print(table)
+    sys.exit(0 if all_ok else 1)
+
+
+if __name__ == "__main__":
+    main()

slos.yml

@@ -0,0 +1,19 @@
+# Example SLO config — replace with your Prometheus queries
+
+prometheus: http://prometheus.nexus.local:9090
+
+slos:
+  - name: "API availability"
+    query: 'sum(rate(http_requests_total{job="nexus-api",code!~"5.."}[5m])) / sum(rate(http_requests_total{job="nexus-api"}[5m]))'
+    target: 0.999
+    window: 30d
+
+  - name: "API latency p99 < 500ms"
+    query: 'histogram_quantile(0.99, sum(rate(http_request_duration_seconds_bucket{job="nexus-api"}[5m])) by (le))'
+    target: 0.5
+    window: 7d
+
+  - name: "Auth service availability"
+    query: 'sum(rate(http_requests_total{job="nexus-auth",code!~"5.."}[5m])) / sum(rate(http_requests_total{job="nexus-auth"}[5m]))'
+    target: 0.9999
+    window: 30d