[Tutorial/Workflow/Doc] Partitions and Nodes Explained

[TinasheMTapera]

🧾 Summary

A brief tutorial to explain nodes, why they are different, and how you should configure your jobs around them

🎯 Learning Goals / Outcomes

What should a reader understand or be able to do after following this tutorial/workflow?

• what is a node
• what node should I default to
• what nodes can I configure for special cases

📦 Required Tools, Packages, or Data

List any software packages, datasets, or computing platforms used (e.g., R, Python, FASRC, GeoPandas, etc.).

• Tools: scontrol show partition [partition name]
• Data source(s): https://docs.rc.fas.harvard.edu/kb/running-jobs/

📄 Outline / Structure

Provide a brief outline of the sections or steps in the document:

1. Introduction
2. Setup
3. Key Steps
4. Conclusion / Next Steps

🔗 Related Materials

Link to related GitHub repos, issues, lab protocols, or previous examples:

• FASRC Docs

🚦Status & Next Steps

What is the current status?

• Proposal
• In Progress
• Ready for Review
• Completed

Any notes on who will write or review the content?

[TinasheMTapera]

Example:

(base) [ttapera@boslogin06 ~]$ scontrol show partition hsph
PartitionName=hsph
   AllowGroups=hsph_users,slurm-admin AllowAccounts=ALL AllowQos=ALL
   AllocNodes=ALL Default=NO QoS=N/A
   DefaultTime=NONE DisableRootJobs=NO ExclusiveUser=NO ExclusiveTopo=NO GraceTime=0 Hidden=NO
   MaxNodes=UNLIMITED MaxTime=3-00:00:00 MinNodes=0 LLN=NO MaxCPUsPerNode=UNLIMITED MaxCPUsPerSocket=UNLIMITED
   Nodes=holy8a[26303-26306,26403-26406,26503-26506,26603-26606,32101-32112,32201-32212,32301-32312,32509-32512]
   PriorityJobFactor=1 PriorityTier=4 RootOnly=NO ReqResv=NO OverSubscribe=NO
   OverTimeLimit=NONE PreemptMode=OFF
   State=UP TotalCPUs=6272 TotalNodes=56 SelectTypeParameters=NONE
   JobDefaults=(null)
   DefMemPerNode=UNLIMITED MaxMemPerNode=UNLIMITED
   TRES=cpu=6272,mem=57748152M,node=56,billing=7710
   TRESBillingWeights=CPU=0.6,Mem=0.07G

(base) [ttapera@boslogin06 ~]$ scontrol show partition test
PartitionName=test
   AllowGroups=cluster_users,cluster_users_2,slurm-admin AllowAccounts=ALL AllowQos=ALL
   AllocNodes=ALL Default=NO QoS=test
   DefaultTime=NONE DisableRootJobs=NO ExclusiveUser=NO ExclusiveTopo=NO GraceTime=0 Hidden=NO
   MaxNodes=UNLIMITED MaxTime=12:00:00 MinNodes=0 LLN=NO MaxCPUsPerNode=UNLIMITED MaxCPUsPerSocket=UNLIMITED
   Nodes=holy8a[24101-24102,24301-24302,24401-24402,24501-24502,24601-24602,26101-26102,26401-26402,26501-26502,26601-26602]
   PriorityJobFactor=1 PriorityTier=5 RootOnly=NO ReqResv=NO OverSubscribe=NO
   OverTimeLimit=NONE PreemptMode=OFF
   State=UP TotalCPUs=2016 TotalNodes=18 SelectTypeParameters=NONE
   JobDefaults=(null)
   DefMemPerNode=UNLIMITED MaxMemPerNode=UNLIMITED
   TRES=cpu=2016,mem=18561906M,node=18
   TRESBillingWeights=CPU=0.0,Mem=0.0G