Redis 中的密钥过期与 TTL

Redis often holds temporary data — sessions, one-time codes, cached responses. Expiration lets a key delete itself after a set time so memory stays lean.

EXPIRE attaches a time-to-live in seconds to an existing key. Once the clock runs out, the key vanishes. See the example below.

Set value and TTL together with SET ... EX — safer than two separate commands, since there's no window where the key has no expiry.

TTL returns the seconds left on a key. It returns -1 if the key has no expiry, and -2 if the key doesn't exist at all.

For finer control, PEXPIRE and PTTL work in milliseconds instead of seconds. The code below sets and reads a millisecond TTL.

EXPIREAT takes a Unix timestamp, so a key can die at an exact wall-clock moment — like precisely at midnight.

PERSIST strips a key's TTL, making it permanent again — handy when temporary data gets promoted to durable.

Redis removes expired keys two ways: lazy deletion checks on access, and active deletion samples keys in a background task and purges expired ones.

Watch out: a plain SET (no EX) overwrites a value and clears its existing TTL. Modifying commands like APPEND or INCR keep the TTL intact.

TTL powers real patterns: auto-expiring sessions, rate-limit counters, one-time passwords and verification tokens, and caches that refresh after N seconds.

Redis can publish an event when a key expires via keyspace notifications, so your app can react. Enable expired-key events, then subscribe. See below.

Test your expiration knowledge.

Recap: EXPIRE and SET ... EX set a TTL, TTL reads it (-1 none, -2 missing), PEXPIRE/EXPIREAT give finer control, PERSIST removes it, plain overwrites clear it.