Tobias Grieger
Tobias Grieger
Upon further reflection, I have to backtrack a little. I think the committed index isn't strictly needed. Say all nodes in a cluster spontaneously reboot. They'll manage to elect a...
As for the applied index, that's a bit more complicated. It's really only the `StateMachine` itself that can tell whether it's applied something or not (it could apply and then...
> To facilitate efficient state machines which handle their own durability, we may want to add a method to StateMachine that Consensus could use to ask the state machine its...
> > it'll (probably) use 0 as the index of the "last append" > > That would be a bug, the master should use the index of the last commit...
Recap of the above is that adding `last_applied_index()` to the state machine interface is a good idea. Implementations which don't persist state just respond with zero (so that they get...
There's this one: https://www.cs.cmu.edu/~dga/papers/leases-socc2014.pdf I had looked at it before and it seemed promising after some adjustments.
I'm gonna do this one.
> I agree, we should be updating the port and ip address on reconnection, if you want to implement it that would be awesome. Ok, happy to do that. >...
> @tschottdorf We do wait! here you can see it waiting. Note the response from the server doesn't come until the entry is commited! Ah, thanks, the part in `advance_commit_index`...
> I might even be willing to argue that changing address/port demands a membership change. What's the argument for that? I would hope to treat that the same as a...