[HTCondor-users] running parallel jobs using HTCondor

[刘岚 <liulan199331@xxxxxxx>]

Hi all,

I'd like to use HTCondor mostly handling parallel universe jobs along with some vanilla jobs. The scale may up to 10K cores or more, and I am wondering if it is better to have multiple dedicated schedulers and all slots to be partitionable in the pool? if it is, does all the dedicated schedulers have all the dedicated resources information in the pool and can use them with the configuration CLAIM_PARTITIONABLE_LEFTOVERS = true? And the manual says that "the default value for configuration variable NEGOTIATOR_PRE_JOB_RANK has changed to prefer to match multi-core jobs to dynamic slots in a best-fit manner", does it in the situation that COMSUPTION_POLICY=true and what does best-fit mean is to select the minimal one from all candidate slots?

Thanks,

Kathy