This was discussed here https://infinispan.zulipchat.com/#narrow/stream/118645-infinispan/topic/client.20retry

Basically, the retry max count can be replaced by retrying up to the number of servers across all nodes. This retry should also only happen when a server is known to go down.

The socket-timeout instead should also be changed to be a time for which the operation will report a timeout exception if it has not responded from a single server.

These changes remove one configuration option and provide timeout with a more meaningful usage.