记一次 Redis 数据库迁移

笔者通过一个 Redis 数据库迁移的例子，介绍了迁移脚本的执行思路。

作者：马文斌，MySQL/Redis 爱好者~

爱可生开源社区出品，原创内容未经授权不得随意使用，转载请联系小编并注明来源。

本文约 500 字，预计阅读需要 2 分钟。

业务背景

最近因业务需要将集群中的 Redis db8 迁移到另一个库中。不需要把全部的 Redis 都迁移过去，只需要迁移 db8 到新服务器的 db15。

大概情况是这样，如图：

当然也探索了一些开源工具看能否实现，最后效果都不太理想，于是自己手撸一个脚本来迁移吧。

migrate_redis.py

以下是脚本内容。

import redis
import time

# 定义 redis1 和 redis 主机信息
redis1_host = '192.168.1.1'
redis1_port = 6579
redis1_db = 8
redis1_password = 'xxxx'

redis2_host = '192.168.1.1'
redis2_port = 6579
redis2_db = 15
redis2_password = 'xxxx'

# 连接 redis1 和 redis2 并验证密码
redis1 = redis.StrictRedis(host=redis1_host, port=redis1_port, db=redis1_db, password=redis1_password)
redis2 = redis.StrictRedis(host=redis2_host, port=redis2_port, db=redis2_db, password=redis2_password)

# 设置每次批量迁移的数据量
batch_size = 1000

# 为进度跟踪初始化变量
keys_processed = 0
start_time = time.time()

# 使用 SCAN 批量获取 key
cursor = '0'
total_keys = len(redis1.keys('*'))

while cursor != 0:
    cursor, keys = redis1.scan(cursor, count=batch_size)
    for key in keys:
        key_data = redis1.dump(key)
        redis2.restore(key, 0, key_data, replace=True)

        keys_processed += 1

        # 每 1000 个 key 打印一次进度
        if keys_processed % batch_size == 0 or keys_processed == total_keys:
            elapsed_time = time.time() - start_time
            keys_per_second = batch_size / elapsed_time
            estimated_remaining_time = (total_keys - keys_processed) / keys_per_second

            print(f"Processed {keys_processed}/{total_keys} keys. "
                  f"Elapsed Time: {elapsed_time:.2f} seconds. "
                  f"Estimated Remaining Time: {estimated_remaining_time:.2f} seconds for the next 1000 keys.")

            # 为下一批次重置变量
            start_time = time.time()

print("Data migration completed.")

输出效果

每隔 1000 个 key 打印一次输出，并评估剩余迁移时间。

Processed 1000/3592 keys. Elapsed Time: 16.46 seconds. Estimated Remaining Time: 42.67 seconds for the next 1000 keys.
Processed 2000/3592 keys. Elapsed Time: 16.96 seconds. Estimated Remaining Time: 27.01 seconds for the next 1000 keys.
Processed 3000/3592 keys. Elapsed Time: 17.03 seconds. Estimated Remaining Time: 10.08 seconds for the next 1000 keys.
Processed 3592/3592 keys. Elapsed Time: 9.81 seconds. Estimated Remaining Time: 0.00 seconds for the next 1000 keys.
Data migration completed.

Process finished with exit code 0

迁移完之后检查

运行 info 命令检查：

源库 3592 个 key，目标库也是 3592 个 key, 迁移完成，收工！