Python小程序扫描清理Redis中的key

场景

项目中很多地方使用Redis，有的用于缓存，有的直接做为存储，有的key设置有过期，有的key没有过期时间。
随着时间增长，Redis存储数据越来越多，消耗内存不断增长；
无论测试或生产环境，总内存是有限的；
有的key可能临时或测试使用的；
于是有了清理Redis key的需求。

Redis命令

查看key个数：
dbsize
info keyspace

查看内存情况：
info memory

通配符扫描key：
SCAN cursor [MATCH pattern] [COUNT count]

Python小程序

Linux服务器一般都自带Python，这里用Python编写2个小程序，分别用于扫描key和扫描并删除key。

扫描key：

#!/usr/bin/env python
# Scan keys in Redis.
# Author: cdfive
from redis import Redis
import time


def RedisScan(host, port, password, db, cursor, pattern, count):
    start_time = time.time()
    client = Redis(host=host, port=port, password=password, db=db)
    counts, other_cursor_counts = 0, 0
    while True:
        cursor, keys = client.scan(cursor, pattern, count)
        length = len(keys)
        if length > 0:
            for key in keys:
                counts += 1
                print("[%s][%s]%s,cursor=%s" % (
                    int(time.time() - start_time), counts, key.decode("utf-8"), cursor))
        else:
            other_cursor_counts += 1
            print("[%s][other_curosr]other_cursor_counts=%s,cursor=%s" % (
                int(time.time() - start_time), other_cursor_counts, cursor))

        if cursor == 0:
            break
    client.close()
    print("[%s]counts=%s,cursor=%s,other_cursor_counts=%s"
          % (int(time.time() - start_time), counts, cursor, other_cursor_counts))


RedisScan("localhost", 6379, "123456", 0, 0, "*xxx*", 1000)

扫描并删除key：

#!/usr/bin/env python
# Scan and delete keys in Redis.
# Author: cdfive
from redis import Redis
import time


def RedisScanDelete(host, port, password, db, cursor, pattern, count, batch_delete_size):
    start_time = time.time()
    client = Redis(host=host, port=port, password=password, db=db)
    counts, other_cursor_counts = 0, 0
    delete_keys = []
    while True:
        cursor, keys = client.scan(cursor, pattern, count)
        length = len(keys)
        if length > 0:
            index = 0
            for key in keys:
                index += 1
                counts += 1
                print("[%s][%s]%s,cursor=%s" % (
                    int(time.time() - start_time), counts, key.decode("utf-8"), cursor))

                delete_keys.append(key)
                if (len(delete_keys) >= batch_delete_size or index >= length) and len(delete_keys) > 0:
                    client.delete(*delete_keys)
                    delete_keys = []
        else:
            other_cursor_counts += 1
            print("[%s][other_curosr]other_cursor_counts=%s,cursor=%s" % (
                int(time.time() - start_time), other_cursor_counts, cursor))

        if cursor == 0:
            break
    client.close()
    print("[%s]counts=%s,cursor=%s,other_cursor_counts=%s"
          % (int(time.time() - start_time), counts, cursor, other_cursor_counts))


RedisScanDelete("localhost", 6379, "123456", 0, 0, "*xxx*", 1000, 100)

注：RedisScanDelete最后1个参数为batch_delete_size，用于通过pipeline批量删除key，提高删除效率。