I’ve announced I will release 2.2.7 or 2.8.0, what ever the new release after 2.2.6, once Linux kernel 4.10 is released but I didn’t do any work on that. Sorry for inconvenience.
The reason is that my work, I mean as a full-time engineer in a startup company, is very busy right now. I have lots to do about the storage software including fixing bugs and designing and implementing new features. I have no any spare time to do Writeboost works.
The business will definitely continue in March. So it will hopefully be in April that I can release the new version. Please wait some while.
Is it possible to use Raft algorithm with state machine that may returns error when applying a log?
I am implementing Raft application but the state machine may need to return error when applying log, which I should then return it to client. If this happens, the result of applying logs may diverse across nodes in the cluster; if some node fails with the log but other doesn’t.
Thanks for asking the question. As you can see, it wasn’t a trivial
question and answering it provoked an interesting discussion!