WAIT

Replication is asynchronous: a write is acknowledged to the client before replicas have it. WAIT lets a client that needs more certainty wait until at least numreplicas replicas have acknowledged every write this connection made so far. It returns as soon as that happens, or when timeout milliseconds pass, and replies with the number of replicas that have the writes either way.

WAIT reduces the window for losing a write when the primary fails, but it does not make replication synchronous: a write that times out is not undone.

The connection is blocked while it waits; other clients are served as usual. Inside MULTI or a script, WAIT replies at once with the current count.

Options

Argument	Meaning
numreplicas	How many replicas must acknowledge.
timeout	Milliseconds to wait at most. 0 waits as long as it takes.

Return value

Case	Reply (RESP2 and RESP3)
On a primary	Integer: replicas that acknowledged this connection’s writes
On a replica	Integer 0, at once
numreplicas is not an integer	Error ERR value is not an integer or out of range
timeout is not an integer	Error ERR timeout is not an integer or out of range
timeout is negative	Error ERR timeout is negative

Examples

127.0.0.1:6379> SET order:1001 "paid"
OK
127.0.0.1:6379> WAIT 1 500
(integer) 1
127.0.0.1:6379> WAIT 3 500
(integer) 2

Compatibility

Same arguments and replies as Redis 7 and Valkey 8 on a primary. On a replica, Redis replies with an error; HotKV replies 0.

See also

WAITAOF, ROLE, INFO (replication), replication

In a cluster: Replicas and failover