SRANDMEMBER key [count]

When called with just the key argument, return a random element from the set value stored at key.

If the provided count argument is positive, return an array of distinct elements. The array's length is either count or the set's cardinality (SCARD), whichever is lower.

If called with a negative count, the behavior changes and the command is allowed to return the same element multiple times. In this case, the number of returned elements is the absolute value of the specified count.

*Return value

Bulk string reply: without the additional count argument, the command returns a Bulk Reply with the randomly selected element, or nil when key does not exist.

Array reply: when the additional count argument is passed, the command returns an array of elements, or an empty array when key does not exist.

*Examples

redis>  SADD myset one two three
(integer) 3
redis>  SRANDMEMBER myset
"three"
redis>  SRANDMEMBER myset 2
1) "one"
2) "two"
redis>  SRANDMEMBER myset -5
1) "one"
2) "three"
3) "one"
4) "three"
5) "one"
redis> 

*History

  • >= 2.6.0: Added the optional count argument.

*Specification of the behavior when count is passed

When the count argument is a positive value this command behaves as follows:

  • No repeated elements are returned.
  • If count is bigger than the set's cardinality, the command will only return the whole set without additional elements.
  • The order of elements in the reply is not truly random, so it is up to the client to shuffle them if needed.

When the count is a negative value, the behavior changes as follows:

  • Repeating elements are possible.
  • Exactly count elements, or an empty array if the set is empty (non-existing key), are always returned.
  • The order of elements in the reply is truly random.

*Distribution of returned elements

Note: this section is relevant only for Redis 5 or below, as Redis 6 implements a fairer algorithm.

The distribution of the returned elements is far from perfect when the number of elements in the set is small, this is due to the fact that we used an approximated random element function that does not really guarantees good distribution.

The algorithm used, that is implemented inside dict.c, samples the hash table buckets to find a non-empty one. Once a non empty bucket is found, since we use chaining in our hash table implementation, the number of elements inside the bucket is checked and a random element is selected.

This means that if you have two non-empty buckets in the entire hash table, and one has three elements while one has just one, the element that is alone in its bucket will be returned with much higher probability.