🪵 Grafana Alloy
Grafana Alloy is our heavy-duty telemetry collector for logs! Think of it as a massive pipeline: it gathers application and system logs from across our Kubernetes clusters and securely funnels them straight into our Loki storage backend.
(Note for image update: Please add an image here showing Alloy's place in the log pipeline: Pods -> Alloy -> Loki).
🏗️ Architecture & Deployment
To keep things ultra-simple and highly scalable, we run Alloy as a Standalone Deployment instead of a DaemonSet.
Alloy ingests logs in two distinct ways:
- 🎯 Direct Ingestion (Primary): It exposes a dedicated internal endpoint, allowing your applications to push logs directly to it over the network. This is our standard, highly recommended approach.
- 🧲 Dynamic Scraping (Fallback): It automatically discovers and reads logs from Kubernetes pods via a custom label. This method is strictly reserved as a fallback for stubborn 3rd-party services that can't push logs directly.
🧹 Built-in Filtering (Noise Reduction!)
Logs get messy fast. To save on expensive storage costs and keep your search queries lightning-fast, Alloy actively intercepts and drops useless logs before they ever hit Loki!
We currently drop logs hitting these endpoints:
healthlivenessmetrics