1

Task queue priority management

I am having a hard time getting my head around Clarity task queue management. I can see tasks enter the queue based on the value of Priority. However, my task servers still appear to claim tasks in the order the task was added to the queue regardless of Priority. My task servers are almost always in continuous use because of data harvesting we do on project models to track model usage and health. Those tasks have a priority of 10. However, teams with critical needs (i.e. print deadlines) create tasks with a priority less than that. I want those to jump the queue line and be serviced next by the task servers. This does not appear to be the case. Am I missing something?

1 reply

Jeff - one thing that might interfere with your plan is that, by default, we do have options to combine tasks together, if there are multiple tasks in the queue for the same target?

The idea is that because file opening can take so long, it's worth it to open the model once, and then perform all of the tasks in the queue, regardless of their priority at that time.

So - it should still be considering the tasks in order of priority... But basically other tasks with lower priority could be "jumping on" when the original high priority task gets claimed.

Does that sound like a possible explanation in this case?

If so: the other option in the Server / Settings / Task Processing Settings  is to set the "Claim All Tasks For Each Model" to FALSE. The down side would be that the model will get opened more than once - but the upside would be that it should be more strict about running tasks in closer to perfect order by priority.

Matt 

JM

Thanks Matt, for the feedback.  I have intentionally turned off the claiming of all tasks for one model because we open models with varying workset conditions.  But that really does not matter to me.  What concerns me is that 1) there appears to be no queue jumping based on priority and 2) important tasks are getting dropped because they were not claimed in a reasonable amount of time.  My task priority timeout is set to 480 minutes.  More than long enough for the data harvesting tools to do their jobs on a model-by-model basis.  I just want the assurance that if I submit a task with a priority of 1, it will be the next task claimed when a task server becomes available regardless of how many tasks are already in the queue.