Plans for Automatic failover of failed node
As I understand it the current version of membase has no type of automatic failover for a failed node. We are considering using membase but having all of the entries that were in the failed node becoming unavailable until someone manually intervenes is not acceptable for us. I have read and understand the reasons why automatic failover is difficult (because it is not always clear what to do), but I was wondering if there are plans for implementing configurable automatic failover?
In our use case we would not necessarily need complete failover. It would be acceptable if only "reads" failed over to some other replica in the cluster. This seems like it would be easier because I think that most of the uncertaintly about automatic failover is due to writes (you don't want to be writing to two separate nodes that think they are both the master for that key). I spoke with a membase person at a meetup in Boulder a couple of months ago, and he said that it might not be too hard to do this read failover maybe at the moxi proxy level, because the client should be able to get a list of available nodes that have a particular key.
Anyway, I was wondering if there are any plans for implementing some sort of automatic failover (or just read failover), and if so if there are any ballpark time frames for this type of feature? I really like what you guys are doing, but the lack of failover may be a deal breaker for us.