Observability - Elestio blog

Sign in Subscribe

Observability

A collection of 6 posts

Self-Host Langfuse to Trace Your AI Agents

Self-Host Langfuse to Trace Your AI Agents

Your agent worked fine in the demo. Then it hit production, and one afternoon a user asked it something ordinary and it burned through 40 tool calls, spent way too many tokens, and returned a confidently wrong answer. You open your logs and find a single line: "agent finished.

Inside Postgres pg_stat_statements: Find Slow Queries Without an APM

Inside Postgres pg_stat_statements: Find Slow Queries Without an APM

Production Postgres falls over for one of three reasons: a missing index, the same query running ten thousand times when it should run ten, or a join that decided today is the day to seq-scan a 50-million-row table. The trick is knowing which one. Without an APM

Promtail Is Dead: How to Migrate Your Log Pipeline to Grafana Alloy Before It Breaks

Promtail Is Dead: How to Migrate Your Log Pipeline to Grafana Alloy Before It Breaks

Promtail hit end-of-life on March 2, 2026. If you're still running it, your log pipeline is now officially unsupported. No more security patches, no more bug fixes, no more updates. Grafana merged Promtail's code into Alloy over a year ago, and the Loki Helm

SigNoz + OpenTelemetry: Build a Complete Observability Stack for Your Microservices

SigNoz + OpenTelemetry: Build a Complete Observability Stack for Your Microservices

Your microservices are slow. Users are complaining. And somewhere in your distributed system, there's a bottleneck you can't find because you're flying blind. Most teams solve this by throwing money at Datadog or New Relic. Then they get the invoice and start questioning their

Grafana + Prometheus + Loki: Build a Complete Observability Stack

Grafana + Prometheus + Loki: Build a Complete Observability Stack

Metrics without logs are useless. Logs without metrics are noise. You need both, correlated, in one place. That's what the Grafana observability stack gives you. The combination of Prometheus (metrics), Loki (logs), and Grafana (visualization) has become the de facto standard for self-hosted observability. It's

Prometheus + Alertmanager: Build Production-Ready Monitoring with Custom Alert Rules

Prometheus + Alertmanager: Build Production-Ready Monitoring with Custom Alert Rules

Your servers are on fire and you find out from a customer tweet. Sound familiar? That's what happens when you rely on uptime checks instead of real monitoring. Prometheus is the monitoring system that powers most of the cloud-native world. Paired with Alertmanager, it becomes a complete