Hello, I'm a PhD student at Princeton and I just started perusing the Zookeeper codebase and wanted to ask about the consistency guarantees of zookeeper.
In the original paper <https://www.usenix.org/legacy/event/atc10/tech/full_papers/Hunt.pdf> from 2011, the claim is that Zookeeper provides A-Linearizability for writes and FIFO client order (which seems like it's talking about sequential consistency/process order) for reads. But in the documentation <https://github.com/apache/zookeeper/blob/master/zookeeper-docs/src/main/resources/markdown/zookeeperOver.md>, which was last updated in 2019, under the "Guarantees" subsection it says the consistency provided is sequential consistency. Now I haven't yet gone into the code in too much detail to look at this from a low level (I'm hoping to in the next few days, so if you also have pointers/tips as to where to look at first that'd be really helpful!), but I just wanted to ask about this discrepancy. What are the consistency guarantees for reads and writes? Is A-Linearizability (as defined in the original paper) still provided? Also thanks for having such a well documented code base! Excited to work with it. Thanks, Anja Kalaba