Print number of keys in Redis

Question

Is there a way to print the number of keys in Redis?

I am aware of

keys *

But that seems slightly heavy weight. - Given that Redis is a key value store maybe this is the only way to do it. But I would still like to see something along the lines of

count keys *

Greengrocer

There's a pull request for COUNT, it got denied though. github.com/antirez/redis/pull/32 antirez also commented on KEYS – Annabell

I wondered if they hadn't supported it as it would be O(n) - guess this confirms it. – Greengrocer

Answer 1

You can issue the INFO command, which returns information and statistics about the server. See here for an example output.

As mentioned in the comments by mVChr, you can use info keyspace directly on the redis-cli.

redis> INFO
# Server
redis_version:6.0.6
redis_git_sha1:00000000
redis_git_dirty:0
redis_build_id:b63575307aaffe0a
redis_mode:standalone
os:Linux 5.4.0-1017-aws x86_64
arch_bits:64
multiplexing_api:epoll
atomicvar_api:atomic-builtin
gcc_version:9.3.0
process_id:2854672
run_id:90a5246f10e0aeb6b02cc2765b485d841ffc924e
tcp_port:6379
uptime_in_seconds:2593097
uptime_in_days:30
hz:10
configured_hz:10
lru_clock:4030200
executable:/usr/local/bin/redis-server

Morbid

redis-cli INFO Keyspace | grep ^db – Horlacher

Answer 2

DBSIZE returns the number of keys and it's easier to parse.

Downside: if a key has expired it may still count.

http://redis.io/commands/dbsize

Isaacisaacs

In that example, KEYS * evicts the expired key. Also Redis may actively evict some expired keys, but not necessarily all of them. – Isaacisaacs

Answer 3

The DBSIZE command returns the number of keys

> DBSIZE

Overpower

Answer 4

WARNING: Do not run this on a production machine.

On a Linux box:

redis-cli KEYS "*" | wc -l

Note: As mentioned in comments below, this is an O(N) operation, so on a large DB with many keys you should not use this. For smaller deployments, it should be fine.

Alfano

Extremely handy, and lets you filter on keys as well. – Saunderson

That's an O(n) operation, is there any way to do this in O(1)? – Platelayer

Do not use on large database in production environment. KEYS Command – Amundson

Somebody is going to read this, do this on a production box some day without thinking it through and then push it over the edge...probably has already happened. – Johannessen

This should have a disclaimer to only use on non-production servers. Otherwise you should use redis.io/commands/SCAN – Friedcake

can confirm: it has happened. – Canea

This answer is outright dangerous and there is no need to use this O(n) operation when INFO KEYSPACE and DBSIZE will do! – Nomarchy

I don't mean to offend but this answer is same as what is explained in the question without the wc -l command and the author is asking for an alternative. – Tymothy

Answer 5

Since Redis 2.6, lua is supported, you can get number of wildcard keys like this

eval "return #redis.call('keys', 'prefix-*')" 0

see eval command

Izak

Using KEYS to count keys (with or without a prefix) is like throwing the baby out with the bathwater. – Sheerness

For Lua newbies: the # in this code is the length operator. – Deka

Answer 6

Go to redis-cli and use below command

info keyspace

It may help someone

Sunder

Answer 7

1. Displays the DB's name along with keys count:

redis-cli info keyspace
# Keyspace
db0:keys=12995,expires=0,avg_ttl=0
db12:keys=5524396,expires=5,avg_ttl=45201

2. Display the no. of keys in the selected database:

redis-cli dbsize
(integer) 12995

3. Displays whole Redis's stats:

redis-cli info

Levitate

Answer 8

dbsize() returns the total number of keys.

You can quickly estimate the number of keys matching a given pattern by sampling keys at random, then checking what fraction of them matches the pattern.

Example in python; counting all keys starting with prefix_:

import redis
r = redis.StrictRedis(host = 'localhost', port=6379)
iter=1000
print 'Approximately', r.dbsize() * float(sum([r.randomkey().startswith('prefix_') for i in xrange(iter)])) / iter

Even iter=100 gives a decent estimate in my case, yet is very fast, compared to keys prefix_.

An improvement is to sample 1000 keys on every request, but keep the total count, so that after two requests you'll divide by 2000, after three requests you'll divide by 3000. Thus, if your application is interested in the total number of matching keys fairly often, then every time it will get closer and closer to the true value.

Cortisone

Answer 9

After Redis 2.6, the result of INFO command are splitted by sections. In the "keyspace" section, there are "keys" and "expired keys" fields to tell how many keys are there.

Shakeup

This is not correct. This is a sample output of the section : # Keyspace db0:keys=366,expires=366 Here, 'keys' indicates total no of keys and 'expires' indicates no of keys with expiry set. Essentially it means that they have a ttl set and they are set to expire, not that they have expired. – Wail

Answer 10

eval "local count = redis.call('scan', 0, 'match', 'key:*:key', 'count', 10000) if count ~= 0 then return #count[2] end " 0

eval "local count = redis.call('sscan', 'key.key:all', 0, 'match', '*', 'count', 1000000) if count ~= 0 then return #count[2] end " 0

Darbies