Our Weblogic persistence uses a cookie insert persistence method. This seems to be what the F5 guide and the BEA Weblogic template uses as well.
It works _most_ of the time. However, we do see entries in the weblogic logs that indicate that from time to time a request comes in and falls on the wrong server (not the primary or secondary weblogic server). That puts weblogic into a recovery state to reconstruct the session on the server that it arrived on.
Sounds very similar to the situation Moose reports above. Thanks for the code snippet. I think I can use it for a variation to log when this occurs and capture some information to try to determine what happened.
I'd also be interested in anyone who has found a solution to this situation. Save me the work of trying figure this one out-- guessing it will take a bit to do that.
Thanks!