SCD suggestions for specifying IBM job resources

Bluesky

SCD software engineers are seeing problems on IBM computers babyblue and blackforest caused by jobs specifying incorrect task information. These jobs probably ran on bluesky's 8-way and 32-way nodes, but are incorrect for babyblue and blackforest 4-way nodes.

Jobs requesting more tasks per node than processors cannot be scheduled, and they prevent scheduling other work until the engineers determine the cause and cancel them. These jobs will not run until the user corrects them.

Please check your LoadLeveler scripts for task and node correctness before submitting them. Check these LoadLeveler keyword combinations:

#@tasks_per_node
#@total_tasks
#@task_geometry
#@node

An incorrect task combination for babyblue and blackforest 4-way nodes is:

total_tasks = 8 & node =1

A correct one is:

total_tasks=8 & node=2

The engineers are studying ideas for trapping and rejecting jobs with invalid task requests, for possible deployment later this year. In the meantime, they thank you for checking your scripts for correctness.