JWT Expiration and Revocation
You'll learn how to control JWT lifetimes and why revoking stateless tokens is surprisingly hard.
Controlling Token Lifetime
JWTs can include time-based claims to limit their validity window:
exp(expiration): A Unix timestamp after which the token is invalid. The server should reject any token where the current time exceedsexp.nbf(not before): A Unix timestamp before which the token shouldn't be accepted yet. Useful for tokens issued ahead of time.
When validating, always check these claims:
// Pseudocode validation
if (currentTime >= token.exp) {
throw "Token expired";
}
if (currentTime < token.nbf) {
throw "Token not yet valid";
}
Why it matters: Short expiration times (minutes to hours) limit damage if a token is stolen. The attacker's window shrinks dramatically.
The Revocation Problem
Here's the catch: JWTs are stateless by design. The server doesn't store them or maintain a session database. It just validates the signature and checks claims. But what if you need to revoke a valid token immediately—like when a user logs out or changes their password?
You can't truly revoke a stateless JWT without adding state back in:
- Maintain a blocklist of revoked token IDs (
jticlaim) until theirexppasses - Store tokens in Redis or similar and check on every request (loses statelessness benefit)
- Keep expiration times very short (5-15 minutes) and use refresh tokens for renewal
Analogy: It's like a movie ticket with a timestamp. Once printed with a valid signature, the theater accepts it—you can't remotely "cancel" that specific ticket unless they check a cancellation list at the door.
Key Takeaway: Use exp claims to limit JWT lifetime automatically, but accept that true revocation requires reintroducing server-side state—making short expiration times your best defense against compromise.