Free to read ยท Three levels

Prometheus: time series, PromQL, and alert contracts

Eighteen developed lessons cover Prometheus collection, series identity, PromQL, histograms, rules, relabeling, storage, routing, and validation.

18 lessonsEnglish + ArabicContent reviewed
Start learning

In this course

Follow a fictional warehouse service from instrument contracts through collection, series, queries, rules, and routing. The course emphasizes meaning: what a counter counts, which population an expression selects, and how identity, windows, absence, and limits change the decision.

Each level contains six complete lessons with explained examples, diagnostic mistakes, worked exercises, and official references. Python programs use local synthetic values only. Every PromQL, YAML, exposition, and flag block is non-executable teaching material; no Prometheus, Alertmanager, target, or receiver is started.

What you will learn

  • Write metric, target, label, unit, and completion-boundary contracts.
  • Interpret PromQL selectors, rates, aggregation, matching, histograms, and rules with their time semantics.
  • Review relabeling, cardinality, storage, copies, routing, and tests while separating models from actual implementation evidence.

Your learning path

Three levels

Basic: collection and time-series meaning

Follow collection boundaries, series identity, metric types, targets, selectors, and counter rates.

  1. The pull model and measurement boundaries
  2. Time series, labels, and identity
  3. Counters, gauges, and exposition contracts
  4. Jobs, targets, and scrape health
  5. Selectors, vectors, and evaluation time
  6. Counter rates, increases, and resets
Explore this level

Intermediate: PromQL and rule contracts

Build compatible aggregation, matching, classic histograms, recording rules, alerts, and relabeling designs.

  1. Aggregation and meaningful error ratios
  2. Vector matching and metadata enrichment
  3. Classic histograms and quantile estimates
  4. Recording rules and derived series contracts
  5. Alert rules and state lifecycles
  6. Discovery and relabeling stages
Explore this level

Advanced: scale, persistence, and safe validation

Review cardinality, absence, local storage, remote copies, Alertmanager, and measurement-change evidence.

  1. Cardinality, churn, and ingestion limits
  2. Staleness, absence, and time comparisons
  3. Local storage, retention, and recovery assumptions
  4. Remote write, federation, and data copies
  5. Alertmanager routing, grouping, and suppression
  6. Rule testing and controlled measurement changes
Explore this level

Who this is for

  • Developers, operations and platform engineers, and learners designing or reviewing Prometheus metrics, queries, and rules.
  • Teams seeking interpretable measurements and reviewable alerts with controlled identity, cost, and evidence boundaries.

Before you start

  • Basic services, HTTP, and programming knowledge; YAML and monitoring familiarity helps but system administration experience is unnecessary.
  • Basic Python reading is sufficient for calculations, with no cloud account, running service, or package installation required.

Related technical solutions

up is zero or an expected target is absentDistinguish discovery loss from a failed scrape and isolate timing, transport, exposition, and limit failures using bounded evidence.A query is empty or vector matching failsTrace selection, sample windows, grouping, and matching keys before broadening a query or forcing a fallback zero.Series growth causes collection or query pressureIdentify label multiplication, churn, target growth, and histogram changes before raising limits or deleting monitoring data.