So wrapping it up, if I get you right you say that:
- there's no way to completely avoid reading stale data
- it has nothing to do with running on a cluster or on a single server
- the application should handle this problem itself, not the O/R mapper
Does that mean that we manually have to call refresh after any operation where I want to avoid potentially stale data? Is there any other patterns to minimize the chance of reading stale data, especially in a clustered environment?
(Also, I might be wrong but I do see a difference between potentally stale data and 100% certain stale data - like reading from a node's cache while the other node got already updated).
I understand that this discussion might get out of scope, but still, I'm very grateful reading your thoughts on this topic.
[Message sent by forum member 'kjozsa' (kjozsa)]
http://forums.java.net/jive/thread.jspa?messageID=237396