anotherJJz

I know a `Clinet` may maintain multiple `Connection`. but just for a `Connection`, its tcp four tuple should be provided so that the user can identify it. It is useful...

I have tried this way, but still only get the remote address of the connection. This is not enough to identify a connection. When i use event to collect metrics...

> VER and ABS should not exists at the same time. I know that, But I want to specify a another version but increment one when the expected version is...

> First of all, OP of #2251 #2252 retracted him/herself from the idea. > > @jjz921024 Is overriding `createSocket()` of JedisSocketFactory an option for you? Where you'd be able to...

> Have you encountered any problems (production-level) caused by this option in the actual environment? there no，except exception stack log `Connection reset by peer` in server side. But, I think...

Hi @git-hulk , May i question? How do we implement this feature? 1. For each column family add a config, specifying its number of write buffers. eg: `pubsub.max_write_buffer_number`、`propagte.max_write_buffer_number`、`zset_score.max_write_buffer_number` ... 2....

Yes, I want to try

> I believe "lite mode" is not a suitable name as it does not clearly convey its purpose. > > We should find a more specific name instead of labeling...

For setting the `write_buff_size` for each column family. There are a simple way that i can think of is to extend `config set rocksdb.write_buffer_size num` command. For example, the command...

@Zheaoli 在这个pr中，原本也实现过在reader目录下增加一个保存元数据的文件。但后面考虑仅就增量同步这个功能，目前需要保存的信息只有`repl_id`和`repl_offset`，想着`repl_id`可以添加到目录上作为区分（目前目录上仅ip和端口作为区分应该不保险）；`repl_offset`记录在aof文件名上（目前aof文件名仅是一个递增的序号，应该没有具体含义？），这种实现有点类似kafka记录msg的方式base_offset + size 能定位到具体某条消息