We are preparing our application for clustered deployment on V 3.2.4. So far everything seems to work fine when we are using commit level A and cache-invalidation: clustering works, invalidation messages are sent and confirmed as can be seen in the locks.
Nevertheless, not all bean instances seem to react on the invalidation messages and stick to their local state. We can come around this (at least in several tries), if we reduce max-bean-age to a low value (like 10). But we get several ctx-lock warnings then.
Those effects make me worry about the confidence I should have in our clustered solution. I checked my configuration several times against the payed documentation but I think that's all fine.
If anyone could point me to possible error sources here or give me a hint how to investigate that further, I would be very pleased.
Volker
