[GitHub] activemq-artemis issue #2480: ARTEMIS-2212 Avoid using CLQ on ServerConsumer...
Github user franz1981 commented on the issue:
> Whats the concurrent performance on this. Do you have some stats?
Only allocations profiles and given that is in the hot path, we have a LOT of ConcurrentLinkedQueue::Node saved by this change.
The overall perf won't change by a bit (if not a tiny improvement), because it isn't a bottleneck AFAIK.
> Re sync lock blocks. Could be replaced with lock on this and or sync methods.
Agree at a first look, but I do not remember the reason why was used a separate lock TBH