JPPF Issue Tracker
star_faded.png
Please log in to bookmark issues
feature_request_small.png
OPEN  Feature request JPPF-429  -  Use existing node connections for heartbeat-based connection checks
Posted Dec 31, 2015 - updated Sep 25, 2017
action_vote_minus_faded.png
0
Votes
action_vote_plus_faded.png
lolo4j (lolocohen) has been working on this issue since March 24, 2016 (23:10)
Issue details
  • Type of issue
    Feature request
  • Status
     
    New
  • Assigned to
     lolo4j
  • Type of bug
    Not triaged
  • Likelihood
    Not triaged
  • Effect
    Not triaged
  • Posted by
     lolo4j
  • Owned by
    Not owned by anyone
  • Category
    Networking
  • Resolution
    Not determined
  • Priority
    Normal
  • Targetted for
    icon_milestones.png JPPF Vision
Issue description
Currently the check for hard failures of node to server connections is done by setting up an additional connection on a separate port, along with the use of a hearbeat mechanism to send short messages and receive an acknowlegement within a limited time frame. The configuration of this mechanism is too complicated, as it requires configuration properties on bith server and node sides. Furthermore, there shouldn't be a need for an additional connection, which impacts the grid performance and scalability.

We propose to implement a mechanism that uses the existing job channel connection to perform the same checks. The mechanism should be suspended while job data is being transfered between the node and ddriver, and resume when the node is either idle or executing the tasks.

There shouldn't be any configuration on the node side, except for enabling/disabling the heartbeat service, and the driver should send any related configuration properties during the handshake.