These are chat archives for atomix/atomix

1st
Apr 2017
Himanshu
@himanshug
Apr 01 2017 02:56
@kuujo the problem mentioned is about removing a crashed node permanently, only option available right now is that , "user sends a request with address of crashed node to us and that initiates server.cluster().member(someAddress).remove() " ... and we are unsure if that works.
so I am trying to do the simpler static configuration model to take away error likelihood. currently its not possible because copycat would store configuration.
eventually I do want to have dynamically evolving cluster , but bringing in a major new dependency has its own set of risks and it might be better to start small and simple.
also, by making it optional we are actually making copycat more general.... then copycat users can make choice on whether/how they want to manage configuration.
that said, I understand your point of view of it being difficult to implement.
Himanshu
@himanshug
Apr 01 2017 03:02
and, I am open to other suggestions as well (after all this is exploration phase) :)