transition from a master-slave Redis setup to a sharded cluster

[Aditya Mishra]

Is it possible to transition from a master-slave Redis setup to a sharded cluster while reusing the same data path for Redis data files?

In the current setup, there are 3 VMs running with one VM serving as the Redis master and the other two VMs acting as slaves. The plan is to create a Redis sharded cluster with new instances on these VMs. The options are to either make each VM a master and add more VMs as slaves or create multiple Redis instances on one VM and then cluster them.

The proposed benefit of using the same data path is to avoid physically moving the data. Instead, the idea is to extract the data from the memory of the VM where it was stored in the master-slave setup and provide it to the sharded cluster for storage or specify the existing path for the new Redis instances in the cluster.

While uncertain about the feasibility of this approach, there is a willingness to try it when migrating from the master-slave to a sharded cluster configuration.

[Sotto Voce]

Can you expand on what you mean when you say "reusing the same data path for Redis data files"?  The reason I ask is "path" is a work usually used when talking about data stored on a filesystem.  The free open-source Redis keeps its data in RAM, where a process can arrange its in-memory data in hierarchical structures, but they aren't often called "paths".

Or are you referring to the filesystem/disk location where Redis saves its data recovery snapshots?