Request processing chain has been improved both in terms of performance and
correctness, more specifically:
1) A lock-free pausable executor has been implemented in order to replace the
old blocking one.
2) Local processing is now executed by the same requesting thread, to avoid
unneeded context switching.
3) When the application is paused, the processor holds queued commands and
schedules them to be later handled by the router, due to possible membership
changes which may cause the command to be rerouted.
Original issue reported on code.google.com by sergio.b...@gmail.com on 11 Sep 2010 at 8:12
Original issue reported on code.google.com by
sergio.b...@gmail.com
on 11 Sep 2010 at 8:12