Operations
Operations covers running your software in production: configuration, building, deploying, environments, GitOps, incidents, Kubernetes, networking, error tracking, logging, metrics, monitoring and tracing. The pages are about the developer's side, such as how you configure your app for Kubernetes and logging. Setting up the cloud platform underneath is in the Cloud Handbook.
- Configuration: configuration per environment in git, overrides, secrets and merge requests.
- Networking: DNS, load balancers and CDNs.
- Production readiness
- Startup scripts: an image entrypoint that runs setup scripts from
/docker-entrypoint.d/before the command, in Docker, Kubernetes and GitLab CI.
Continuous integration
- Building: pipeline stages and job dependencies.
- Versions: recording which build runs in the error tracker, a Kubernetes label, a
build_infometric and image annotations, and keeping it out of public response headers.
Continuous delivery
- Deploying: deploys with Helm and promotion between environments.
- Deploys: pushing from CI with a token, runners and agents in the cluster, and GitOps controllers that pull.
- Environments: production, staging, development, testing and preview environments.
- GitOps: the repository a controller deploys from, promotion, rollback and drift.
Incidents
- Response: when to declare an incident, the procedure, roles, severity levels, mitigation and communication.
- Reports: writing a blameless incident report and holding the incident report meeting.
- Template: an incident report template to copy.
- Examples: a finished incident report, based on a real database incident.
Kubernetes
- Apply: client-side and server-side apply.
- Autoscaling: the Horizontal Pod Autoscaler.
- Deployments: rolling updates, shutdown, readiness, liveness and startup probes, and what Kubernetes does when a container, pod or node fails.
- Scheduling: how the scheduler filters and scores nodes.
- Services: DNS and endpoints of a Service.
Observability
- Error tracking: the Sentry SDK, releases, environments and notifications.
- Logging: which data goes in the logs and which in the database, metrics or cache, log levels from debug to critical, and trace IDs.
- Metrics: counters, gauges, histograms, summaries and exemplars.
- Monitoring: alerts and dashboards.
- Tracing: spans, propagation and sampling.