Reliability

Reliability is about controlled behaviour under failure. A reliable Norscode service does not need to avoid every error. It needs to fail in ways that are bounded, understandable and recoverable for both users and operators.

Timeouts, retries and fallbacks

Timeouts prevent one slow dependency from consuming unlimited attention. Retries should be bounded and used only where duplicate effects are safe enough. Fallbacks should be explicit and tested, not assumed.

funksjon kan_prøve_igjen(status: heltall) -> bool {
    hvis status == 429 eller status == 500 eller status == 503 da {
        returner true
    }
    returner false
}

Incident flow

Useful incident handling usually follows the same pattern: detect, isolate, compensate, patch, then learn. Reliability improves most when the team reduces uncertainty about what should happen first during an active issue.

Read this alongside logging, observability and deploy for the full operational picture.