Prometheus cheat sheet

First success: README. Definitions: glossary.

Run (Docker)

cd examples/docker
docker compose up -d
# UI: http://localhost:9090
# Targets: Status → Targets
docker compose logs prometheus
docker compose down

Reload config after editing prometheus.yml (compose enables lifecycle):

curl -X POST http://localhost:9090/-/reload

Objects

Object	Role
Target	Something that answers `GET /metrics`
Job	A named scrape config (`job_name`)
Series	Metric name + labels
Scrape	One pull
TSDB	On-disk samples
PromQL	Questions over the TSDB
Alert	Rule that stays true; sent to Alertmanager
Exporter	`/metrics` adapter for something you did not write
ServiceMonitor	Operator CRD: scrape these Services

Metric types

Type	Goes	Typical query
Counter	Up (or reset)	`rate(metric[5m])`
Gauge	Up and down	`metric`
Histogram	Buckets + sum + count	`histogram_quantile(0.99, sum(rate(metric_bucket[5m])) by (le))`
Summary	Client quantiles	Do not `avg` across instances

PromQL you will type

up
prometheus_build_info
rate(prometheus_http_requests_total[5m])
up{job="prometheus"}
sum(rate(http_requests_total{status=~"5.."}[5m])) by (job)
  / sum(rate(http_requests_total[5m])) by (job)

rate() needs a range vector ([5m]). Filter with {label="value"}. Match regex with =~.

Files in this module

File	What it is
examples/docker/docker-compose.yml	Laptop Prometheus
examples/docker/prometheus.yml	Scrape config
examples/prometheus-rule.yaml	Operator `PrometheusRule` (HighErrorRate)
examples/alertmanager-config.yaml	Alertmanager routes to Slack
examples/servicemonitor.yaml	Operator scrape of `app: my-app`

Production flags (not first success)

--storage.tsdb.retention.time=15d
scrape_interval: 15s
# never: user_id, request_id, raw email as labels

Helm (cluster, after Docker works): prometheus-community/kube-prometheus-stack.