KBI 311566 Issue Addressed: Tasks Might Be Skipped When Relator Is Dynamically Scheduled With Short Intervals And Associated With Large Amount Of Monitored Nodes
Version
Argent Advanced Technology 5.1A-1707-B or below
Date
Monday, 21 August 2017
Summary
When a Relator is dynamically scheduled, the associated Nodes are monitored evenly within scheduled intervals.
For example, a Relator running once every 10 minutes should be scheduled to run at 10:00, 10:10, 10:20 etc, and load is distributed within each 10-minute interval.
If the system is under heavy load, the Supervising Engine might take a bit longer to schedule tasks.
Let’s say the Supervising Engine is 30 seconds late and recalculates at 10:10:30, it will attempt to skip Nodes that should have been scheduled in the 30-second period.
This becomes serious when the Relator has hundreds of Nodes scheduled every minute.
The issue has been addressed in Argent Advanced Technology 5.1A-1707-C or above
Technical Background
N/A
Resolution
Upgrade to Argent Advanced Technology 5.1A-1707-C or above
For customers who encounter such an issue and cannot upgrade immediately, one workaround is to break the Relator with a lot of associated Nodes into multiple Relators, and distribute Nodes among them.