This sure looks like a bug...?
bos at mailrank.com
Mon Apr 18 20:08:57 EDT 2011
On Mon, Apr 18, 2011 at 4:59 PM, Dan Reverri <dan at basho.com> wrote:
> I've tried to provide a walk through below that explains the behavior. The
> main lesson to take away is you should always provide a client id and vector
Thanks for the description of what's going on. From your description, it
looks like this is a class of bug that could be very difficult for clients
to program defensively against. The case of a missing vclock is easy enough
(flag a conflict against whatever was returned by the server), but what if a
client erroneously issues two successive puts to the same bucket+key with
the same vclock? Since clients have to treat vclocks as opaque, I can't
think of a way to identify the case of "this put was rejected because the
vclock was considered stale".
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the riak-users