Re: [HTCondor-users] Arbitrarily limit the number of dynamic slots

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

On 1/18/2023 9:06 AM, Charles Goyard wrote:

> Hi all,
>
>
> I just started deploying dynamic slots on my pool (Debian, HTCondor 10),
> and it works well.  However, I would like to set an arbitrary limit on
> the number of slots than can be created on a single machine.
>
> This is because our sweet spot is somewhere around 6 to 8 jobs running
> in parallel.
>
> I could use predefined, static slots to the same effect, but I was
> wondering if there is a way to keep dynamic slots and state : "this
> machine can be cut in 6 at most".
>
> My setup is very simple for dynamic slots, as it is only :
>
> use feature:PartitionableSlot
>     
    
    Hi Charles,
    
    To achieve the above desired policy, you could leverage the fact
    that all slots have an attribute "TotalSlots" which is the number of
    slots being advertised by that startd.  So perhaps just a startd
    START _expression_ to look at this, keeping in mind that the
    partitionable slot itself counts as one, so to have 6 jobs at most
    running you want TotalSlots kept to be at 7 or below.
    
    So you could drop the following into your config  (and do a
    condor_reconfig or restart of the startd):
# Setup up the Execution Point (EP, i.e. the startd) to use dynamic slots, but
# never run more than 6 jobs at most (no matter how the EP is carved up).
use feature:PartitionableSlot
START = $(START) && ( TotalSlots <= 7 ) 
    
    How did I come up with the above?  I just did a "condor_status -l"
    on a partitionable slot machine and looked at the piles of
    attributes I could reference :).  Instead of a strict scalar
    comparison, you could config it to be a function taking into account
    any attribute in the slot ad (number of cores, type of processors,
    amount of RAM, whatever), and/or you can customize it for each and
    every server if you wanted.  Documentation about the "START" config
    knob is at:
    
      
https://htcondor.readthedocs.io/en/latest/admin-manual/policy-configuration.html#the-start-_expression_
    
    but basically the START config knob is a classad _expression_ that
    must evaluate to True in order for the startd to allow a new job
    activation.
    
    
    Hope the above helps,
    regards,
    Todd T
    
    --
Todd Tannenbaum <tannenba@xxxxxxxxxxx>  University of Wisconsin-Madison
Center for High Throughput Computing    Department of Computer Sciences
Calendar: https://tinyurl.com/yd55mtgd  1210 W. Dayton St. Rm #4257