This content has been marked as final. Show 2 replies
what you say makes perfectly sense: each request takes a long time to be processed because of the CPU bottleneck, hence the connection is held for a long time.
I would suggest profiling your application with yourkit or similar tool, to understand where all that CPU is wasted.
Also make sure you have enough heap to handle all that load, monitoring GC time can also be useful.
You might want to throttle requests (eg with a WorkManager) to avoid clogging your system with too much load pushing it beyond the point of maximum throughput.
I identified some methods that have high cpu usage ...and as well i am thinking on putting a work manager until the dev team will fix he issue in the code.
This are some specific methods on a WS.
It is possible to assign work manager (with max constraint) to specific methods (to this level of granularity) ?
If yes , let's say that i put a max thread constraint to 10 what happend to the 11 th concurent request ? it is denied or put on wait ?
(as i remember is denied)