SCAN - Upstash Documentation

Use SCAN to walk through the keys of the database incrementally, a batch at a time. Each call takes a cursor and returns the next cursor together with a batch of keys. Start with cursor 0 and keep calling with the cursor from the previous reply until the server returns 0 again, which marks the end of the iteration. Because the work is split over many short calls, SCAN never blocks the server the way KEYS can on a large keyspace. MATCH filters the returned keys with a glob-style pattern, COUNT hints at how much work each call should do (a hint about effort, not a page size, so batches vary in length), and TYPE limits the reply to keys of one type. Filtering is applied after a batch has been read, so a call can legitimately return no keys at all while the cursor is still non-zero: only the cursor tells you when the iteration is over. The guarantee is that every key present for the whole iteration is returned at least once. Keys added or removed while the scan runs may or may not show up, and a key can be returned more than once, so make the processing of each key idempotent. HSCAN, SSCAN, and ZSCAN apply the same mechanism inside a single collection.

Syntax

SCAN <cursor> [MATCH <pattern>] [COUNT <count>] [TYPE <type>]

Arguments

Argument	Required	Repeatable	Description
`<cursor>`	Yes	No	Cursor returned by the previous call; start at `0`.
`MATCH <pattern>`	No	No	Return only elements matching this glob-style pattern.
`COUNT <count>`	No	No	Hint for how much work each iteration should do.
`TYPE <type>`	No	No	Return only keys of this type, such as `string`, `list`, or `hash`.

Important points

This operation can inspect a large part of the database. Prefer cursor-based scans where possible and avoid unbounded use on hot paths.
The cursor is opaque. Start with 0 and continue until the server returns cursor 0; a single iteration may return no elements.

Response

The reply reports the result of the operation. Error replies have the same shape in RESP2 and RESP3 and are surfaced as exceptions by the SDKs below.

Protocol	Reply
RESP2	Two-element array: cursor and array of bulk-string keys
RESP3	Two-element array: cursor and array of bulk-string keys

Client libraries often decode bulk strings, maps, sets, and numeric strings into language-native values. The table describes the Redis wire reply.

Examples

TCP examples use the TLS REDIS_URL from the Upstash console. REST examples use UPSTASH_REDIS_REST_URL and UPSTASH_REDIS_REST_TOKEN.

Redis CLI

SCAN 0

@upstash/redis

import { Redis } from "@upstash/redis";

const redis = Redis.fromEnv();

const [cursor, keys] = await redis.scan(0, { match: "*" });

upstash_redis

from upstash_redis import Redis

redis = Redis.from_env()
result = redis.scan(0)
print(result)

ioredis

import Redis from "ioredis";

const redis = new Redis(process.env.REDIS_URL!);
const result = await redis.scan("0");
console.log(result);

node-redis

import { createClient } from "redis";

const client = await createClient({ url: process.env.REDIS_URL })
  .on("error", console.error)
  .connect();
const result = await client.scan("0");
console.log(result);

redis-py

import os
import redis

client = redis.from_url(os.environ["REDIS_URL"])
result = client.scan(0)
print(result)

go-redis

package main

import (
    "context"
    "fmt"
    "os"

    "github.com/redis/go-redis/v9"
)

func main() {
    opts, err := redis.ParseURL(os.Getenv("REDIS_URL"))
    if err != nil {
        panic(err)
    }
    client := redis.NewClient(opts)
    result, _, err := client.Scan(context.Background(), 0, "*", 0).Result()
    if err != nil {
        panic(err)
    }
    fmt.Println(result)
}

jedis

import java.net.URI;

import redis.clients.jedis.Jedis;

try (Jedis jedis = new Jedis(new URI(System.getenv("REDIS_URL")))) {
  Object result = jedis.scan("0");
  System.out.println(result);
}

redis-rs

use redis::TypedCommands;

fn main() -> redis::RedisResult<()> {
    let url = std::env::var("REDIS_URL").expect("REDIS_URL is not set");
    let client = redis::Client::open(url)?;
    let mut connection = client.get_connection()?;

    let iter: redis::Iter<String> = connection.scan()?;
    for key in iter {
        println!("{key}");
    }
    Ok(())
}

​Syntax

​Arguments

​Important points

​Response

​Examples

Syntax

Arguments

Important points

Response

Examples