make sure messages are persisted to disk before being consumed on the other
>>> side, it just makes failure recovery 100x easier (unless you just don't
>>> care about missing messages).
>>>
>>>
>>> On Thursday, August 11, 2016 at 7:00:21 AM UTC-6, Eli
I am writing a system that needs to be distributed across various data
centers e.g. US, Europe, Asia. I need to replicate a small amount of
configuration data between the nodes in each datacenter (there are a small
number of nodes in each data center, 1 or 2) and this configuration can be
