Principal investigators (PIs) and Co-Investigators (Co-I) on the NSF MRI (Award#: 1724891 - 2018) grant and the NIH (Grant#: S10OD024973 - 2019) grant are provided access to a priority boost for their group's jobs.  This is managed by using a QOS (quality of service) value in your job script or when launching an interactive app in OnDemand.




Priority boost QOS values include:


supporters - provided to PIs that either include CCR in grants or pay an annual fee - more details


nih - provided to PI groups on the 2019 NIH grant

nihsupport - provided to PI groups on the 2019 NIH grant AND who are also supporters
NOTE: Any PI with active NIH funding may request access to this priority boost.  PIs MUST enter grant information in ColdFront and have an active allocation to the UB-HPC cluster.  Please contact CCR help after this is complete as providing the access to the qos is a manual process.   After verification of the funding, CCR will provide access to the QOS for the priority boost.  Instructions for adding grant information to ColdFront can be found HERE.  


mri - provided to PI groups on the 2018 NSF MRI grant

mrisupport - provided to PI groups on the 2018 NSF MRI grant AND who are also supporters




Where can I use these priority boosts?


supporters - can be used on any partition in the ub-hpc cluster


nih/nihsupport - can be used only on the cascade partition in the ub-hpc cluster


mri/mrisupport - can be used only on the skylake, largemem and gpu partitions in the ub-hpc cluster




How can I tell if I have access to one of these priority boost QOS values?


To check to see if you have access to any of the priority boost QOS values, look at your SLURM associations:


slimits


You will see a column for QOS.  In this example, the first account has the standard  QOS values available to all users of the ub-hpc cluster:




The example below shows an account with access to the 'supporters' qos: 



If you have access to any of these grant QOS boosts, you'll see either nih/nihsupport and/or mri/mrisupport values listed under the QOS section of the ub-hpc cluster.



 You will also see these listed as an allocation under your project in ColdFront:


NOTE: The resource type is listed as 'cluster partition' which is a bit misleading.  This is done so the proper mapping happens on the systems.




How do I utilize the priority boost?


CCR uses the SLURM Quality of Service (QOS) function for providing group's with an extra boost in priority for their jobs. To take advantage of your quality of service priority boost, please add the following to your SLURM scripts. NOTE: only one qos can be used at a time so groups that are supporters and also have access to a grant QOS should use the combination of the two (i.e. mrisupport or nihsupport)

#SBATCH --qos=supporters


or


#SBATCH --qos=nih


or 

#SBATCH --qos=mrisupport



In OnDemand, enter one of these QOS values in the QOS field, depending on which partition you've selected:





Why isn't my job running first?

Please note that a priority boost is not a ticket to the front of the line (queue).  It is one of multiple factors that go into calculating a job's priority.  Please see this article for more information on Slurm's multi-factor scheduling.




Renewing the allocation for a priority boost:


Users and PIs will get notification emails when the priority boost allocation is set to expire.  You can NOT renew these yourself.  The expiration dates for the priority boosts provided as part of the 2019 NIH S10 award and the 2017 NSF MRI award are set by CCR and based on when the equipment went into production, not when the grant started.  Meaning - we've given you more time so we will not extend these priority boosts past the set expiration date.


If you also have access to the 'mrisupport' or 'nihsupport' QOS values, they will expire at the same time as your 'supporters' boost.  Users in your group can continue to use 'mri' and 'nih' QOS values until they expired.