The default configuration for modern HTCondor is to set up partitionable slots. As each job comes in, the requested number of CPUs and megabytes are allocated
to a dynamic slot in which the job runs. When it finishes, the slot’s resources are returned to the partitionable slot. If you run:
condor_status –af name slottype
… it will show you the slot names and types. The “slot1” is the partitionable slot, and slot1_1, slot1_2, etc. are the dynamic slots.
There’s a config setting to revert to the old behavior of one core per slot and evenly-divided memory, but I don’t recall what it is off the top of my head and
the ReadTheDocs site is throwing a certificate error at my proxy at the moment so I can’t look it up. But it’s there.
Michael V Pelletier
Digital Transormation & Innovation
Currently I have a pool where each workstation is listed as a single slot. Instinctively this feels like a sub-optimal use of resources but I don't know how to change the allocation of slots to be
1 core per slot. Are there any best practice recommendations for how to define slots on a pool?