Given that now we have a single interface for all the connection modes, single, sentinel, cluster and replication, perhaps a good contribution would be a wrapper helper that implements opinionated recovery semantics. For example, wrap the client, and register an exception handler that catches recoverable errors. In such events, queue all calls to "send" and "batch" until the underlying client is replaced with a new one.
Use cases
This will make the vertx redis client even easier to be used by end-users.
Read me
Read this first before creating an issue:
Describe the feature
This is suggested by @pmlopes in https://groups.google.com/g/vertx/c/vZZOFtFGOlI.
To quote his suggestion:
Use cases
This will make the vertx redis client even easier to be used by end-users.
Contribution
Not decided yet.