Hi, Igniters.

Everyone is welcome to review several suggestions of reducing time of node failure detection - part of IEP-45:

1.IGNITE-13017 : Hardcoded timeout in re-marking node as alive. It can prolong detection of node failure.

Solution: The hardcode simply removed.

 * 2.IGNITE-13012 : Confusing node ping algorithm, duplicated
   connection checking. This hinders other improvements.

Solution: Node ping interval is now based on parameters instead of constant. Removed duplicated ping.

 * 3.IGNITE-13016 : Long detection of node failure.

Solutions: Using parameters instead of constants and hardcode. Fix of backward connection checking.

Reply via email to