Aller au contenu principal
OPS

OBSERVABILITY STACK Agent

Deployment of a complete observability stack (Prometheus, Grafana, Loki, Alertmanager).

Request context

<arguments>

Objective

Deploy a production-ready observability stack, with Docker Compose for dev/staging or Kubernetes with Helm for production.

Workflow

  • Identify the deployment mode (Docker Compose, Kubernetes, Victoria Metrics, managed)
  • Generate the configuration files structure
  • Configure Prometheus (scrape configs, alert rules)
  • Configure Grafana (provisioning datasources, dashboards)
  • Configure Alertmanager (routes, Slack/PagerDuty/email receivers)
  • Configure Loki + Promtail (log aggregation)
  • Add node-exporter and cAdvisor for system metrics
  • Document the deployment and verification commands

Expected output

  1. Complete configuration: docker-compose.yml or Helm values
  2. Prometheus: prometheus.yml, alert.rules.yml
  3. Alertmanager: alertmanager.yml with receivers
  4. Loki/Promtail: configs for log aggregation
  5. Grafana: datasources and dashboards provisioning
AgentUsage
/ops:ops-monitoringInstrument application code
/ops:ops-k8sDeploy on Kubernetes
/ops:ops-dockerContainerize the services

IMPORTANT: Always test the stack in staging before production.

IMPORTANT: Configure alerts BEFORE deploying to production.

YOU MUST have persistent storage for the metrics data.

NEVER expose Prometheus/Alertmanager without authentication in production.


See also