Hi,
I am yet to get authorization to share production configuration, But as a developer pretty eager to learn the GFE way of solving this issue. I forsee that same siutation could happen when entries are expired using Quartz kind of scheduler exactly same time on replicated region.
There are two different regions - Region1 and Region2.
t1 ----> Cache Entries are entered to replicated Region-1 (and replicated to Region-2)
t2 ----> Scheduler marks entry to expire at some time in future on Region-1.
t3 ----> Expiry attributes are replicated to Region-2.
t4 ----> Cache entry expired on Region-1 (and Region-2 at the same time).
t5 ----> Expiry replicated to Region-1 => Region-2 (and Region-2 => Region-1)
t6 ----> EntryDestroyedException are handled and logged by GFE.
Hope that explains the issue. I am interested in design (and pattern) to solve this kind of problem. Not seeking for specifc to our current production. If GFE is smart, may I know how it handles? If not, Can I intercept the expiry replication (at t5), ignore them smartly (if I am not overlooking anything)?
Regards