I have not used distributed ones in actual development. Let’s discuss it with you.
My thoughts are:

1. Use redis Pub/Sub method to subscribe to the updates of the secondary cache. When a machine updates the cache, thenPub一个Key,通知所有的客户端，客户端前往更新，刷新一级缓存。key的值可以按机器号:数据key,这样搞，通过机器号判断让更新者不要更新，其它机器更新，数据keyupdate the specified data.
2. Consider whether the first-level cache is really needed. Will not using the first-level cache cause any problems? If it is a performance issue that can be ignored, I tend to use redis directly as the first-level cache, saving time, effort and complexity :)

The background scheduled task synchronizes the redis cache to the memory cache