ReadMeFeed / Community knowledge
Real questions. Useful conversations. Find the people who know your stack.
DevOps & observability · #prometheus
Recording rules precompute PromQL expressions into stored time series to speed up dashboards, but they increase series count and storage cost. Learn when to apply them.
Pre-aggregate expensive PromQL queries into new time series so dashboards query a single metric instead of recomputing range vectors on every refresh.
Configure Prometheus remote write to avoid OOM, certificate outages, and sample loss. Practical steps for buffer sizing, TLS rotation, and error monitoring.
Turn slow Prometheus dashboards into snappy, pre‑computed views by using recording rules. Learn how to write, validate, and deploy rules, and understand the trade‑offs that come with pre‑computing metrics.
Stop relying on manual UI edits for your Grafana dashboards. Learn how to use file-based provisioning to implement a version-controlled, code-first monitoring workflow.
Dropwizard’s health‑check framework lets you write custom checks, secure the /healthcheck endpoint, and export results to Prometheus while avoiding blocking threads and data leaks.
Sample loss in Prometheus remote write is usually backpressure. Sizing queue shards, batch size and retry backoff lets a growing cluster absorb bursts without unbounded memory growth.
Use metric_relabel_configs to drop high-cardinality labels before samples reach TSDB. A keep rule plus labeldrop for path reduces series count and query cost without changing the exporter.
Dropwizard ships with a rich metrics library that can be exposed as Prometheus data out of the box. This guide walks through enabling the built‑in metrics bundle, registering a custom counter, and verifying the output—plus a quick look at the performance trade‑offs.
grafana
dropwizard
kubernetes