If join/refresh operations fail for some reason on the remote target of the RPC, we need to make sure not to do things at the RPC source (i.e., process trying to join or refresh a group) that should only be done on success (like modify buffer sizes, ultimately leading to segfault mentioned in #64).
If join/refresh operations fail for some reason on the remote target of the RPC, we need to make sure not to do things at the RPC source (i.e., process trying to join or refresh a group) that should only be done on success (like modify buffer sizes, ultimately leading to segfault mentioned in #64).
Fixes #64.