Details
-
Enhancement
-
Resolution: Unresolved
-
Major
-
None
-
None
Description
Today we have a custom netty Connection pool that works great under normal circumstances. However, under certain edge cases (namely sockets being kill in the middle of processing bytes) it can leave some commands in an inconsistent state causing eventual timeouts.
The pool itself never seems to get in a bad state, but commands timing out instead of retrying or relaying an exception immediately is much more preferred.
Once the HR server is confirmed to have proper pipelining support, including consolidating flushing, we should be able to remove this pool as our server is fully non blocking. This would reduce the code complexity of the client by a good chunk.
Attachments
Issue Links
- is related to
-
ISPN-14808 HotRod backpressure support
- New