Keep serving when the flag service is down

How to do it

A flag service is called on almost every request, so its outage becomes yours unless the client survives it. Use a client that downloads the rules and decides locally, as Unleash server SDKs and many OpenFeature providers do. Set a short timeout on the download, as Set timeouts on every call says. Choose each default so a missing value keeps today's behaviour, as Separate deploying from releasing says. Flag files in the repository follow Roll out settings and data files like code.

Examples

TypeScript example

const flags = new FlagClient({ url: FLAG_URL, refreshIntervalMs: 15_000, timeoutMs: 1_000, cacheFile: '/var/cache/app/flags.json' });

export const isNewCheckoutOn = (user: User) => flags.isEnabled('new-checkout', { user, defaultValue: false });

Python example

flags = FlagClient(url=FLAG_URL, refresh_interval=15, timeout=1.0, cache_file="/var/cache/app/flags.json")

def is_new_checkout_on(user: User) -> bool:
    return flags.is_enabled("new-checkout", user=user, default=False)

How to check

Block the flag service on staging and restart the app. Every request still succeeds, and each flag returns its last known value or its default.

Sources

• OpenFeature specification: Flag evaluation

Related rules

• API-35 Set timeouts on every call in API design
• ENG-25 Separate deploying from releasing in Engineering
• ENG-12 Roll out settings and data files like code in Engineering