How to handle stale reads in Riak for things like sessions in the face of network partition -

if 1 storing sessions in riak, there possibility of read being stale.

isn't bad? example, session data. user logs out , refreshes page , still shown logged in.

also, if there network partition or making node unresponsive (excessive i/o or cpu usage), user log out destroys session if subsequent read happens on other side of nodes still going logged in because on side session hasn't been destroyed. answer?

with consistent datastore riak there risk of stale reads under conditions in environments offset riak's availability profile (where stale reads better no reads or inability write).

riak have features can session storage expiry (when using bitcask or in memory backends) , riak documents have examples , links real world session store use cases built on riak worth look: http://docs.basho.com/riak/latest/dev/data-modeling/.